ContentsInformation TechnologyAn Enhanced Scheduling Scheme in IEEE 802.11e HCCA Wireless Networks ………………………………………………………………………………Li FENG, Jianqing LI (1)MedicineDetection and Fatty Acid Analysis of Public sites Legionella Isolated from Macau …………… Hongbo ZHAO,Wenda GUAN, Zifeng YANG, Xi LU, Lei SHI, FeiCHOI, Ziyao MO (8)Application of Disjoint and Conjoint Hard Models for Classification of Different Samples using Chemometric Pattern Recognition …………………………………… Dong LI, Wai-Kei LAM (16)Management and TourismAnalysis on the Effects of Leadership Behavior Organizational Performance ……………… Chao LIU (26)Some Insights into the Development of Museums in Macau: a Perspective of Post-museum Era ………………………………………………………………………… Juan TANG, Yuzhu LIU (39)A Study on Customer Satisfaction of Macau Food Carnival …………………………………Xin WANG (48)Estimating China's Interest Rate Risk Based on Asymmetric GARCH-VaR Models: An Empirical Analysis of Shibor ………………………………………………………………Hao DING, Chuan PANG (58)A Study on Grand Prix Development Strategies Based on the Community Participation of the Macao Residents ………………………………………………………………………… Wai Long LEI (68)Pedagogica ExperienceCultivating Online Self-regulated Learning for Adult Learners in Hong Kong: Its Barriers and Countermeasures ……………………………………………………… Aliana Man Wai LEONG (82)Humanities and ArtsOriginal Culture Research of God of Gambling Belief of China …………………………………………………………………… Zichao WANG, Keling WANG (91)Duanshu Wang and The Preliminary Selection of Ladies＇ Poems ……………………… Yan ZHANG (98)LawReposition and Construction of Victim-Offender Mediation:Protection of Victims Rights under Restorative Justice Pattern ……………………………………………………………………… Bo PENG (104)
1第 5 卷 第 1 期 澳 門 科 技 大 學 學 報 Vol.5 No.12011 年 6 月 30 日 Journal of Macau University of Science and Technology June 30 , 2011 An Enhanced Scheduling Scheme in IEEE 802.11e HCCA Wireless NetworksLi FENG1,2, Jianqing LI1*(1.Faculty of Information Technology, Macau University of Science and Technology, Macao, China) (2.Department of General Studies, Macau University of Science and Technology, Macao, China)Abstract: To improve bandwidth utilization, a recent scheme called the equal-spacing (equal-SP) design was developed to enhance the reference design introduced in IEEE 802.11e HCCA wireless networks. The equal-SP design can remarkably improve the bandwidth utilization of the reference design by lengthening scheduling spacings of different streams. This paper proposes an equal-SP-δ design to improve the equal-SP design further. Compared with the equal-SP design, the equal-SP-δ design introduces a δ-factor in the TXOP calculation and schedules different real-time streams with longer spacings; numerical results show that it has a potential of improving bandwidth utilization, up to 8% or more. Besides, the equal-SP-δ design might reduce the additional computational complexity introduced in the equal-SP design.Key words: HCCA; QoS; Equal-SP; Bandwidth utilizationIEEE 802.11e HCCA無線網絡中一種改進的調度策略馮 麗1，2，李建慶1*（1.澳門科技大學資訊科技學院，澳門；2.澳門科技大學通識教育部，澳門）摘要：為了提高帶寬利用率，最近提出的基於等間隔的調度策略改進了IEEE 802.11e 標準中的參考調度策略。等間隔設計的主要思想是通過增長調度間隔來顯著提高參考調度策略的帶寬利用率。本文對等間隔設計進行改進，提出等間隔δ設計。和等間隔設計相比，等間隔δ設計在發送機會的計算方法中引入了一個δ因子，使之能用更長的間隔調度不同的實時流；數值結果顯示等間隔δ設計在等間隔設計的基礎上進一步提高帶寬利用率8%以上。此外，等間隔δ設計還可以減少由等間隔設計引入的計算複雜性。關鍵詞：混合控制信道訪問方式；服務質量；等間隔；帶寬利用率Received 01 Mar. 2011; Revised 21 Mar. 2011.Project: Science and Technology Development Fund, Macao SAR under Grant No.016/2010/A.Corresponding author: Jianqing LI, Male, Associate professor of Macau University of Science and Technology, his research interests are computer network architecture, network protocol, network security, ITS and so on. E-mail:jqli@must.edu.mo, Tel: (853)88972194.
2Introduction0 IEEE 802.11e[1] is a recent standard that supports quality-of-service (QoS) for wireless LAN applications. In the standard, a reference design consisting of a scheduler and an admission control is provided for guaranteeing the delay requirement of real-time applications. In the reference design, when a new stream requests admission, the scheduler first calculates a common service interval (SI), being equal to the minimum among all delay bound of streams, for all streams, then calculates the transmission opportunity (TXOP) duration for each stream by the SI and the traffic specification (TSPEC) parameters (such as the mean data rate and mean packet size), and next the admission control unit decides whether to admit the new stream. Once the new stream is admitted, the scheduler periodically polls each stream over the SI in a round robin fashion. However, the reference design often overallocates the bandwidth since it uses the most stringent delay bound for all streams. In [2], an equal-spacing (equal-SP) design is developed to generalize and improve the reference design. In the equal-SP design, the spacing of a stream is defined as the interval between two successive schedules of the stream. The equal-SP scheduling means that the scheduler always schedules each stream with equal spacing, while it may schedule different streams with different spacings. In contrast, the reference scheduler schedules all streams with a common minimum spacing. Compared with the reference design, the equal-SP design has two advantages: (a) it has a potential to improve bandwidth utilization, up to 22% or more, and (b) it can reduce the polling overheads significantly. However, there is a limitation in the equal-SP design: the scheduling spacing generated by the delay bound is required to be harmonic. That is, a longer spacing must be divisible by a shorter spacing. This means that a longer delay bound may possibly be shorten significantly, which will limit the improvement of the bandwidth utilization badly. This paper aims at improving the equal-SP design further. To this end, this paper proposes a new design - the equal-SP-δ design. Besides the improvement of bandwidth utilization, the new design might reduce additional computational complexity in the equal-SP design. The basic idea behind the equal-SP-δ design is to loosen the constraint of generating harmonic spacing and resort to longer spacings. Instead of generating harmonic spacing, the equal-SP-δ design requires the integer-multiple spacings (i.e., the longer spacing must be an integer multiple of the smallest spacing), and at the same time, it introduces a δ-factor in the TXOP calculation of the equal-SP design. The goal is to ensure that the bandwidth calculated by the new TXOP calculation method must benefit from longer spacings. Numerical results show that compared with the equal-SP design, the equal-SP-δ design has a potential to improve bandwidth utilization, up to 8% or more. The rest of the paper is organized as follows. Section 1 outlines the equal-SP design and explains its limitation. Section 2 proposes the equal-SP-δ design. Section 3 presents numerical results. Section 4 concludes the paper. The equal-SP design1 This section presents the equal-SP design in [2] and then points out its drawbacks.The equal-SP design111 The equal-SP design, consisting of a scheduler and an admission control unit, is outlined here. Its basic idea is to schedule different streams with different spacings, rather than schedule all streams in a common spacing, as in the reference design[1].The scheduler The scheduler uses the mandatory set of TSPEC parameters of streams to generate a scheduling: mean data rate, nominal MAC service data unit (MSDU) size, and maximum SI or delay bound. Consider n admitted streams with SI1≤SI2≤…≤SIn. The scheduling for the n streams is executed in four steps.Step 1: the scheduler generates a set of harmonic spacings by . Here, we let denote the harmonic spacing of stream i and is required to satisfy the two requirements: (a)
3Li FENG,et al. An Enhanced Scheduling Scheme in IEEE 802.11e HCCA Wireless Networksand (b) , for , where denotes is divisible by .Step 2: the scheduler updates in step 1, where in the updated is calculated as follows: (1)In (1) , ρ, Li, Ri, and M denote the mean data rate, nominal MSDU size, negotiated minimum physical (PHY) rate, and maximum allowable size of MSDUs for stream i, respectively; O is the per-packet overheads of the packet transmission, including the transmission time of MAC header and CRC (TmacHeader), PLCP header and Preamble (TphyHeader), Interframe Space (TSIFS), and ACK frame (TACK); P is the per-station polling overhead, i.e., the time spent on SIFS and the transmission time of a CF-Poll frame (TCFPoll); denotes the smallest integer not less than x; denotes the bandwidth allocated to stream i if its scheduling spacing is x; and denotes the normalized bandwidth occupied by stream i. Note that since there are possibly multiple spacings satisfying the requirements (a) and (b) in step 1, the min() function in (1) ensures that the chosen spacing is the best among the generated spacings in step 1 in terms of bandwidth utilization. In addition, in the definition of , denotes the estimated number of MSDUs of stream arrived during [0,x]; denotes the per-packet transmission time of stream i; denotes the transmission time of maximum allowable size of MSDUs of stream i; the max() function in (1)ensures that at least one packet can be transmitted for each TXOP duration.Step 3: the scheduler calculates for , where and is defined in (1).S t e p 4 : t h e s c h e d u l e r s c h e d u l e s by the ERM algorithm[2], where and are from Steps 2 and 3, respectively. According to the EMR algorithm, stream i is scheduled with once for each scheduling spacing ; the scheduling of a stream with a longer spacing can be preempted by a stream with a shorter spacing; the scheduling of all streams is required to execute continuously unless a preemption happens. Fig. 1(a) illustrates an example of the equal-SP scheduling when the harmonic spacing set for three admitted streams with , where the time unit is millisecond (ms).LEGEND:(a)(b)TXOP 1TXOP 2TXOP 3h1T =25 msh3T =50 msh2T =50 msm1T =25 msm2T =50 msm2T =75 msFig. 1 An example of the equal-SP scheduling when (a) the harmonic spacing set and (b) the integer-multiple spacing set for 3 admitted streams with , where the time unit is ms.The admission control unit When a new stream requests admission, the admission control unit accepts the new stream if , where is the normalized bandwidth allocated to HCCA traffic, and is the normalized bandwidth occupied by streams; in which is defined by (2)In (2), n-1 is the number of existing streams, n is used as the index for the newly arriving stream, , and , where and are defined in (1) .Limitation of the equal-SP design111 The equal-SP design schedules streams with longer by adopting longer spacing. The advantages of adopting long spacing are two folds: (a) it can improve the bandwidth utilization and (b) it can reduce the polling overheads. For example, assume that the deadline of a stream is 50 ms. If the stream is scheduled every 50 ms, the allocated TXOP is 15
4ms while if it is scheduled every 25 ms, the allocated TXOP is 10 ms, then it is better to be scheduled every 50 ms than every 25 ms because the normalized bandwidth is 15/50 in the former which is less than 10/25 in the later and at the same time the scheduler needs to poll the stream once in the former while twice in the later. The equal-SP design implies that the longer the scheduling spacing, the larger the potential of improving bandwidth utilization. However, the equal-SP design requires the scheduling spacing to be harmonic. That is, the longer spacing should be divisible by the short spacing, i.e., , which may shorten the feasible scheduling spacing. For example, for three admitted streams with , the harmonic spacing set is and Fig. 1(a) illustrates the scheduling of the three streams. One possible method to generate the longer spacing than the harmonic spacing requires that the spacing of stream i is equal to the integer-multiple of stream 1’s spacing. That is, denotes the integer-multiple spacing of stream i and let , where denotes the largest integer not greater than x, and . In the example, the integer-multiple spacing set is and Fig. 1(b) illustrates the scheduling of the three streams. It can be shown from Fig. 1 that the scheduling spacing of TXOP 3 is lengthened from 50 ms for the harmonic spacing (Fig. 1(a)) to 75 ms for the integer-multiple spacing (Fig. 1(b)), therefore the bandwidth utilization using integer-multiple spacings can be improved. In Section 2, we will propose a new design called the equal-SP-δ design, utilizing integer-multiple spacing, to improve the equal-SP design further.The equal-SP-δ design1 This section presents the equal-SP-δ to tackle with the drawback described in Section 1.2. The proposed design uses the four mandatory TSPEC parameters of streams, as in the equal-SP design, However, there are three changes: (a) generates the integer-multiple spacing, rather than the harmonic spacing; (b) generates the TXOP by using a function different from in (1); (c) requires B ≤0.693 in this paper. Among the three changes, point (b) can avoid addition computational complexity in the equal-SP design, which is explained in the Step 2 in the equal-SP-δ design, while points (a) and (c) are explained as follows.Point (a) is used to generate longer spacing for the same stream, consequently having greater potential of improving the bandwidth utilization. The reason is that the integer-multiple spacing of stream i, , is the largest integer-multiple of , where . Therefore , where the harmonic spacing of stream i, , satisfy and .Point (c) is involved with the schedulability of stream i. The scheduling scheme in HCCA networks requires that (i) all admitted streams should be schedulable and (ii) the normalized bandwidth occupied by the admitted streams should not be larger than the normalized bandwidth allocated to HCCA traffic, B.Recall that the equal-SP scheduling requires to generate harmonic spacings, i.e., a longer spacing is divisible by a shorter spacing. Theorem 1 in [2] proved that with harmonic spacings, all admitted streams are schedulable as long as the sum of their normalized bandwidths, , is not larger than 100%, i.e., . Since ensures necessarily, the admission control unit in the equal-SP design only requires .HCCA is designed to coexist with EDCA in IEEE 802.11e. The percentage of bandwidth allocated to HCCA traffic, B, is not specified in IEEE 802.11e standard but its typical value is possibly far less than 100%. For example, B is set to 0.4 in [3]. Then, our idea is that if B is far less than 1, we can loosen the constraints to generate harmonic spacing and resort to longer spacings, thus both guaranteeing the requirements (i) and (ii) for the scheduling scheme in HCCA networks and improving the bandwidth utilization. To this end, we propose to adopt integer-multiple spacings. With integer-multiple spacings, the schedulability of admitted streams is ensured
5Li FENG,et al. An Enhanced Scheduling Scheme in IEEE 802.11e HCCA Wireless Networksby the Liu and Layland (LL) condition proved in [4]: a set of periodic tasks for any period or spacing is schedulable if . N o t e that the LL condition holds for any positive spacing. We here require generating integer-multiple spacings because we want to avoid the issue of introducing too many fragments, as explained in [2]. Therefore, applying the LL condition, we ensure that all admitted streams with integer-multiple spacings must be schedulable if . The limit of can be always removed at the expense of additional computational complexity. One method is the harmonic chain method[5] in which periods of tasks are formatted into multiple harmonic chains; and in each harmonic chain, a larger period is divisible by a short period. For example, the period set {3, 8, 12, 16, 48} can be organized into two harmonic chains:{8,16} and {3,12,48}, as illustrated in Fig. 2. The 48 in Fig. 2 is called the common root of the two harmonic chains. It was proved in [5] that a task set must be schedulable if the periods of the task set can be formatted into multiple harmonic chains with a common root and B≤1. In contrast, the harmonic spacing method in the equal-SP design requires that all periods should form the only harmonic chain. For example, the period set {3, 8, 12, 16, 48} can be formatted into one harmonic chain {3, 3, 12, 12, 48} in which the original periods 8 and 16 are shortened to 3 and 12, respectively. Therefore, the harmonic chain method[5] generalizes the harmonic spacing method in the equal-SP design.Since we must be able to construct multiple harmonic chains with a common root from original periods by utilizing some methods such as the scheduling period generation method in [2], the schedulable condition, B≤1, holds, hence removing the limit of . The additional computational complexity incurs in constructing such multiple harmonic chains with a common root. However, as a return, the harmonic chain method may potentially enlarge the scheduling spacings of streams and thereby improve the bandwidth utilization further.48123168Fig. 2 The harmonic chains in the task set {3, 8, 12, 16, 48}The equal-SP-δ design111 The equal-SP-δ design, composed of a scheduler and an admission control unit, aims at improving the bandwidth utilization of the equal-SP design and better utilizing the longer SI.The scheduler Consider n admitted streams with . The scheduling for the n streams is executed in three steps.Step 1: the scheduler generates a set of integer-multiple spacings by , where the integer-multiple spacing of stream i, , is generated as follows: (3)Step 2: the scheduler calculates for , where and is defined by (4)where is an empirical value and is set to 0.2 in this paper and other parameters are the same as those in (1). In Step 2, we use , instead of in (1). That is, we use to approximate without introducing significant errors, since always overestimates the actual traffic amount while . Moreover, the slight modification enables to have a good property: is decreasing with respect to x, which implies that adopting the longer spacing can better utilize bandwidth, and thus reducing the computational complexity further since we need not to update the spacing, as in Step 2 in the equal-SP design. In contrast, is not of such a good property and the equal-SP scheduling needs to update the spacing in (1).S t e p 3 : t h e s c h e d u l e r s c h e d u l e s
6 by the ERM algorithm[2], where and are from Steps 1 and 2, respectively.The admission control unit 111 When a new stream requests admission, the admission control unit accepts the new stream if , where is the normalized bandwidth allocated to HCCA traffic, and is the normalized bandwidth occupied by streams; in which the function is defined in (2); and , where and are defined in (3) and (4), respectively. Numerical analysis3 T h i s s e c t i o n m a i n l y d e m o n s t r a t e s t h e effectiveness of the equal-SP-δ design. Here, Table 1 shows the default PHY and MAC parameter setting for HCCA networks. Table 1 PHY and MAC Parameter setting for HCCA networksParameters ValuesM 2304 bytesB 0.6931PHY Rate (R) 12 Mbpsmin PHY Rate (Rmin) 6 MbpsSIFS 10 usmacHeader (26+4) bytes @RphyHeader (20+4) bytes @RminACK (20+4+14) bytes @RminCFPoll 36 bytes @Rminδ 0.2Table 2 The TSPEC parameter setting for 4 streamsType (i=I, II, III, IV) I II III IVSIi (ms) 50 350 650 700ρi (kbps) 8.4 770 60 70Li (byte) 60 1536 1230 1521Ri (Mbps) 54 54 9 9Table 2 defines four types of traffic with the corresponding TSPEC parameter values and each type of traffic is assumed to have one stream. in row 2 of Table 2 denotes the delay bound of traffic i. To show the advantage of our design at different , we vary from 700 ms to 1300 ms while keeping other parameter values unchanged. From the delay bound of the four streams, one of the harmonic spacing sets is , which is used in our numerical examples of Fig. 3 in spite of the change of . In addition, to quantify the benefit of our design, we calculate the average improvement. The improvement of y over x is defined as for a single operating point ; for multiple operating points, such as in our numerical examples in which varies from 700 ms to 1300 ms and other parameter remains unchanged, we calculate the average improvement.0.7 0.8 0.9 1 1.1 1.2 1.30.0470.0480.0490.050.0510.0520.053the total normalized bandwidthS I4 (unit: se cond) Bh [2]Bh*BdFig. 3 The total normalized bandwidth as varies for the equal-SP design in [2] (i.e., the curve), the equal-SP design with integer-multiple spacing (i.e., the curve), and the equal-SP-δ design (i.e., the curve).Fig. 3 plots the total normalized bandwidth as varies for the equal-SP design in [2] (i.e., the curve), the equal-SP design with integer-multiple spacing (i.e., the curve), and the equal-SP-δ design (i.e., the curve), where , , and , while , , and are defined in (1), (4), (3), and (2), respectively. It is shown from Fig. 3 that as varies from 700 ms to 1300 ms, is almost same and is about 0.0527, fluctuates, and is decreasing, but and are less than 0.0527. These results imply that (a) adopting the integer-multiple spacing has a greater potential to improve bandwidth utilization and (b) can guarantee that longer spacings are always better than short spacings while cannot do. Fig. 3 also shows that as varies from 700 ms to 1300 ms, the average improvement of over is 7.95% while
7Li FENG,et al. An Enhanced Scheduling Scheme in IEEE 802.11e HCCA Wireless Networksthe average improvement of over is 8.30%.Conclusions4 This paper proposes the equal-SP-δ design to further improve equal-SP design in IEEE 802.11e HCCA networks. The proposed equal-SP-δ design not only keeps all advantages of the equal-SP design, but also can utilize the bandwidth more efficiently. Numerical results show that the equal-SP-δ design has a potential of improving bandwidth utilization, up to 8% or more. Besides, the equal-SP-δ design might reduce the additional computational complexity introduced in the equal-SP design.ReferencesIEEE 802.11 WG. IEEE Std 802.11e-2005 (Amendment to IEEE Std [1] 802.11, 1999 Edition (Reaff 2003)), Sept. 2005.Zhao Q L, Danny H K Tsang. An Equal-Spacing-Based Design for [2] QoS Guarantee in IEEE 802.11e HCCA Wireless Networks. IEEE Transactions on Mobile Computing, 2008, 7(12): 1474-1490.Mihaela van der Schaar, Yiannis Andreopoulos, Hu Z P. Optimized [3] Scalable Video Streaming over IEEE 802.11a/e HCCA Wireless Networks under Delay Constraints. IEEE Transactions on Mobile Computing, 2006, 5(6): 755-768.Liu, C L and Layland W. Scheduling algorithms for multiprogramming [4] in a hard real-time environment. Journal of the ACM, 1973, 20(1):46-61.Kuo, T and Mok, A. Load adjustment in adaptive real-time systems. [5] Proceedings of the 12th IEEE Real-Time Systems Symposium, Dec 1991, San Antonio, TX, USA. IEEE Computer Society. 160-170.國際交流澳門科技大學成功召開月球化學元素與礦物學國際研討會由澳門科技大學太空科學研究所承辦的月球化學元素與礦物學國際研討會1月13～14日在澳門成功召開。來自美國、日本、法國，中國大陸和臺灣地區的知名月球科學家出席並作學術報告。本世紀人類掀起了新的探月熱潮，月球的化學元素與礦物組成成為了世界各國探月的重要主題。一方面，月球表面物質組成是研究月球起源和演化歷史的重要資料；另一方面，獲得月表礦物、元素的含量和分佈對於未來月球基地選址、建設方案設計和月球資源的開發利用研究具有重要的指導意義。此次國際研討會內容涉及月球的起源、月球表面元素和礦物的分佈、月球的地形地貌與物質成份的關係、月球表面元素遙感反演方法研究等。來自日本Kaguya、中國嫦娥一號和印度Chandrayaan－1 探月衛星的最新科研成果也在會議上進行了展示。日本Kaguya探月計劃首席科學家M. Kato教授做了題為“日本月球探測最新科學成果”的主題報告；國際著名月球科學家、美國Brown大學的Carle Pieters教授做了題為“月球：揭示類地行星前十億年演化歷史的鑰匙”的主題報告。在專題討論會上，與會科學家就月球起源與演化、當前世界各國探月情況、下一步月球科學研究趨勢等內容進行了熱烈的討論，同時也對澳門科技大學在月球科學探測領域的研究成果給予了高度評價。澳門科技大學與美國肯特州州立大學簽訂合作協議為擴大與海外高校的交流合作，提升大學國際化程度，澳門科技大學副校長張曙光教授帶領的管理團隊一行於4月8至14日赴美國著名高校肯特州州立大學(Kent State University)訪問交流。雙方就兩校教師和學生互相交流以及部分專業直接升讀研究生課程簽署了合作協議，亦共同探討了未來在研究生層面合作辦學之可能。訪問期間，張副校長一行與美國肯特州州立大學校領導、各學院院長、教師們及學生們分別進行了共八場交流。張曙光副校長表示是次澳科大與美國肯特州州立大學簽署協議，推動了澳科大面向國際教育市場的進程。澳科大能與肯特州立這所美國國家級一類綜合研究型大學建立學術聯繫和學生培養的合作關係，不僅增加了澳科大師生與國際先進大學的交流機會，也可讓澳科大的師資隊伍及科研人才更趨國際化，從而配合大學“再上臺階”的發展需要。
8第 5 卷 第 1 期 澳 門 科 技 大 學 學 報 Vol.5 No.12011 年 6 月 30 日 Journal of Macau University of Science and Technology June 30 , 2011 收稿日期：2011-1-19；修訂日期：2011-3-16資金資助：澳門科學技術發展基金（039/2007/A3）﹡通訊作者：莫自耀，男，博士，教授，主要研究方向為病原微生物。E-mail: moziyao@163.net，Tel: （853）88972165 澳門地區公共場所軍團菌的檢出和脂肪酸分型研究趙紅波2，關文達2，楊子峰1，2，魯 曦3，石 磊3，蔡 菲4，莫自耀1，2*（1. 澳門科技大學，澳門 ；2. 呼吸疾病國家重點實驗室（廣州醫學院），廣州） （3. 華南理工大學，廣州 ；4. 澳門大學，澳門）摘要：背景：軍團菌引發軍團菌病和Pontiac熱，但有關澳門地區公共場所軍團菌的檢出和分佈鮮見報道。目的：對澳門地區公共場所的軍團菌（又名退伍軍人菌）進行檢測和分型研究。方法：將水樣處理後進行分離培養和血清學鑒定，並利用螢光定量PCR及LAMP(環介導等溫擴增)方法進行快速檢測，同時用脂肪酸分型對所分離的菌株進行鑒定和同源性分析。結果：43份水樣中有25份軍團菌陽性，陽性率為58.1%。分屬8種軍團菌，半數以上為嗜肺軍團菌。多個樣品中可同時分離出不同軍團菌，兩種快速檢測方法與傳統方法鑒定結果相符。嗜肺軍團菌A群在澳門三個區內均有分佈，B群主要分佈在澳門半島和氹仔，C群則主要分佈於澳門半島；其他非嗜肺軍團菌株大多集中分佈在地理位置較近的區域。結論：澳門地區公共場所軍團菌陽性檢出率較高，基本以均勻分散的形式分佈。關鍵詞：軍團菌；澳門；快速檢測；脂肪酸分型Detection and Fatty Acid Analysis of Public sites Legionella Isolated from MacauHongbo ZHAO2,Wenda GUAN2, Zifeng YANG1, 2, Xi LU3, Lei SHI3, Fei4 CHOI, Ziyao MO1, 2*( 1. Macau University of Science and Technology, Macao , China )( 2. State Key Laboratory of Respiratory Diseases (Guangzhou Medical University), Guangzhou , China )( 3. College of Light Industry and Food Technology, South China University of Technology, Guangzhou , China )( 4. University of Macau, Macao, China )Abstract: Background: Legionella pneumophila is the agent of Legionnaires’ disease and Pontiac fever, but little is known about its detection and distribution in the public areas of Macau. Objectives:To detect and to categorize Legionella in environmental water isolated from the pubic area of Macau. Method:Water samples filtered were cultured and serotyped, while the real-time PCR assay and LAMP (loop-mediated isothermal amplification) were applied to detect Legionella rapidly, and fatty acid analysis was also used to identify and cluster the Legionella strains. Results: There were 25 of 43 water samples detected with Legionella, which belonged to eight kinds of Legionella, and the positive rate was 58.1%. Half of them were Legionella pneumophila. Different Legionella strains could be found in one water sample. Both real-time PCR and LAMP obtained almost the same positive results compared with that of the traditional cultivation. Legionella pneumophila cluster A were scattered in three regions of Macau, cluster B were mainly distributed in Macau peninsula and Tampa, and cluster C mainly distributed in Macau peninsula. The rest of other non-Legionella pneumophila were distributed in regions close to each other. Conclusion: The positive rate of Legionella in Macau pubic sites is higher than that of other areas, results from fatty acid analysis indicates that Legionella spreaded well all over Macau.Keywords: Legionella; Macau; Quick detection; Fatty acid analysis
15趙紅波，等 澳門地區公共場所軍團菌的檢出和脂肪酸分型研究有較高的應用價值。本研究中脂肪酸分型的鑒定結果與培養及血清學鑒定結果基本相符，除三株軍團菌血清學鑒定為LP14型而脂肪酸鑒定為Legionella rubrilucens和Legionella shakespearei。可能是由於此兩種軍團菌與LP14型存在血清學交叉反應，或菌株生長情況影響了脂肪酸分型的鑒定結果。F群中4-1和4-2為同一冷卻塔採樣分離得到的菌株，可近似認為是同一株菌，這兩株菌與A100724002為同一冷卻塔不同時期採樣分離得到的菌株，可認為是同一亞型，這也說明脂肪酸分型方法準確可靠，且具有較好的重複性。同樣，該技術也可對菌株的流行趨勢進行分析，如A群中10-1和10-2為同一冷卻塔分離得到的菌株，分析後可認為是同一株菌。不同分離地點的軍團菌也可分為同一群，如B群中的A100724004和A100807001分別採自醫院與一相鄰建築物的冷卻塔，可認為是同一亞型，推測其為同一來源；又如E群中的22-2和36-3來自相距較遠的不同地點但可認為是同一株菌，證明其可能為同一來源，原因可能是含軍團菌的氣溶膠經風吹傳播或由於空調維修等人為因素導致不同地點冷卻塔水中分離得到的菌株為同一來源。從圖1的結果看，所分離的嗜肺軍團菌共分為三群，其中A群在三島內均有分佈，B群主要分佈在澳門半島和氹仔島，C則主要分佈於澳門半島。其他非嗜肺軍團菌各個種可分別分為一群，且各群中菌株基本分佈在地理位置較近的區域，如D群和J群，分別集中分佈於澳門半島內相鄰地區；也有散在分佈於各區域的，如F群，在澳門半島和氹仔島不同地區均有分佈。基本反應出澳門地區各公共場所軍團菌均勻分散的分佈狀態。綜上所述，澳門地區空調系統水樣及噴水池水檢出軍團菌，其污染情況及高檢出率值得重視，未來研究可進一步進行相關的快速檢測及分型研究。參 考 文 獻Fields BS, Benson RF, Besser RE. Legionella and Legionnaires＇ [1] disease: 25 years of investigation. Clin Microbiol Rev, 2002, 15(3): 506-526. Carratala` J and Garcia-Vidal C. An update on Legionella. Curr Opin [2] in Infect Dis, 2010, 23(2): 152–157. 中國新聞網. (2010-12-30) [2010-05-27]. http://www.chinanews.com.[3] cn/ga/ga-sszqf/news/2010/05-27/2307613.shtml. Bonetta Sa, Bonetta Si, Ferretti E, et al. Evaluation of Legionella [4] pneumophila contamination in Italian hotel water systems by quantitative real-time PCR and culture methods. J Appl Microbiol. 2010, 108(5): 1576-1583. Newton HJ, Ang DKY, van Driel IR, et al. Molecular pathogenesis [5] of infections caused by Legionella pneumophila. Clin Microbiol Rev, 2010, 23(2): 274-298. Templeton KE , Scheltinga SA , Sillekens P,et al. Development and [6] clinical evaluation of an internally controlled, single-tube multiplex real-time PCR assay for detection of Legionella pneumophila and other Legionella species. J Clin Microbiol, 2003, 41(9): 4016-4021. Tomita N, Mori Y, Kanda H, et al. Loop-mediated isothermal [7] amplification (LAMP) of gene sequences and simple visual detection of products. Nat Protoc,2008, 3(5): 877 – 882. Njiru ZK, Mikosza AS, Matovu E, et al. African rypanosomiasis: [8] Sensitive and rapid detection of the sub-genus Trypanozoon by loop-mediated isothermal amplification (LAMP) of parasite DNA. Int J Parasitol, 2008,38(5): 589–599. Costa J, Tiago I, da Costa MS, et al. Presence and persistence of [9] Legionella spp. in groundwater. Appl Environ Microbiol, 2005, 71(2): 663–671. Männisto MK, Salkinoja-Salonen MS, Puhakka JA. In situ [10] polychlorophenol bioremediation potential of the indigenous bacterial community of boreal groundwater. Water Res. 2001, 35(10): 2496-2504.張麗霞, 劉凡. 空調冷卻水中軍團菌污染現狀及預防控制. 環境與[11] 健康雜誌, 2009, 26(5): 460-462. 朱佩雲, 陳悅, 沈建民, 等. 上海部分地鐵站空調冷卻塔水軍團菌污[12] 染狀況的調查與分析. 環境與職業醫學, 2002, 19( 5) : 313- 314.劉美真, 陳經雕, 譚海玲, 等. 廣東部分地區空調冷卻塔水軍團菌污[13] 染狀況調查. 華南預防醫學, 2008, 34(1): 65-66. 陸廣智，陳琦，張秋平，等. 珠海市部分集中空調通風系統軍團[14] 菌污染調查. 中國熱帶醫學, 2009，9(4): 781-782.張紅梅，楊希存，霍長友，等. 秦皇島市不同水體中軍團菌污染[15] 狀況調查. 環境與健康雜誌, 2009, 26(6): 506.澳門立法會. 第2/2004號法律，傳染病防治法. 2004. [16] 中華人民共和國衛生部. 公共場所集中空調通風系統衛生規範. [17] 2003. Gobin I, Newton PR, Hartland EL, et al. Infections caused by [18] nonpneumophila species of Legionella. Rev Med Microbiol, 2009, 20(1): 1-11.
16第 5 卷 第 1 期 澳 門 科 技 大 學 學 報 Vol.5 No.12011 年 6 月 30 日 Journal of Macau University of Science and Technology June 30 , 2011 Application of Disjoint and Conjoint Hard Models for Classification of Different Samples using Chemometric Pattern RecognitionDong LI, Wai-Kei LAM( Macau Institute for Applied Research in Medicine and Health, Macau University of Science and Technology, Macao, China )Abstract: Modelling which plays an important role in classification applied for chemometric pattern recognition involves extracting relationship between interesting properties of samples (class memberships of samples) and analytical data (measurements of samples) using statistical and mathematical methods. Because classification involves determining whether unknown samples can be assigned to one or more predefined groupings according to the models, unknown samples may be misclassified and converse conclusion may be drawn unless proper models are formed prior to classification. In order to determine feasible and stable models for different studies, disjoint and conjoint hard Principal Component (PC) modelling (disjoint and conjoint modelling) which require samples to be uniquely assigned to a specific class (or group) are described and estimated in this paper. Disjoint modelling involves developing independent PC models for each class (‘in group’) and all ‘in group’ models are combined together using statistical methods, whereas conjoint model perform Principal Component Analysis (PCA) on the entire dataset so as to form a model. The number of PCs for each model is optimised using 200 bootstraps. A joint decision function combined by D statistic (one-class Quadratic Discriminant Analysis) and Q statistic (square prediction error) is regarded as disjoint classifier to classify each sample in disjoint study, whereas multi-class Quadratic Discriminant Analysis (QDA) is applied as conjoint classifier. Model performance is assessed on 100 test sets using different iterative splits and a couple of performance indices are used to estimate the applicability and stability of the models. The methods are illustrated using two different datasets including 293 samples from nine polymeric groups obtained using thermal profiling and 75 samples from three types of pharmaceutical tablets obtained using LC-MS profiling. For the classification of polymeric samples, disjoint model is more feasible and stable for the separation of different polymeric groups and the prediction of unknown polymers with higher %PA (Predictive ability) and %CR (Classification Rate). For the classification of tablet samples, the %PA and %CR show that the separation of different types of tablets are clear using both models although disjoint model also has higher %PA and %CR. Comparing with disjoint model and conjoint model, disjoint model has advantage of forming flexible and stable models to group unknown samples to one of known classes that may have very different structures, whereas conjoint model is more useful to examine the differences between classes as it is simpler and faster.Key words: Modelling; Disjoint and Conjoint Hard PC models; Classification; Quadratic Discriminant Analysis; Pattern Recognition; ChemometricsReceived 11 Nov. 2011; Revised 06 May 2011.Corresponding auther: Dong Li, PhD, Assistant Chemist, E-mail: ld99031305@msn.com, Tel: +853-8897 2636
17Dong LI,et al. Application of Disjoint and Conjoint Hard Models for Classification of Different Samples using Chemometric Pattern Recognition化學計量學中單獨硬模和聯合硬模在識別不同樣品中的應用李 棟，林偉基（澳門科技大學藥物及健康應用研究所，澳門）摘要：模型化是應用統計學和數學的方法來分析樣品數據和樣品類別之間關係的技術，它在化學計量學的模式識別特別是樣品分類中起著重要的作用。因為樣品的分類是根據模型來對未知樣品進行歸類，所以不恰當的模型可能導致未知樣品的錯誤分類和產生錯誤的結論。為了給不同的樣品分類確定合適和穩定的模型，在本文中講述和評估單獨模化和聯合模化這兩種技術。單獨模化是單獨給每一類樣品建模然後把所有的獨立模型進行合併和用聯合判定函數用於對樣品進行分類，而聯合模化是對所有類型的樣品進行模化和用二元判別分析法對樣品進行分類。由於潛在變量的應用，因此重複的輔助程序用於優化潛在變量的數量。為了評估單獨模型和聯合模型的應用性和穩定性，已知樣品被隨機分為訓練樣品和測試樣品分別用於建立模型和測試模型。單獨模型或者聯合模型的總體性能是由隨機產生的100個模型的平均性能所決定的。模型的具體應用性和穩定性是用不同的性能指數表達出來。在本文中，聚合物樣品和藥片樣品用於舉例說明各種模化方法，推測能力和分類率指數用於評價模型的性能。比較不同的模型，獨立模型更加靈活穩定和適用於不同結構樣品的分類，而聯合模型能更簡單快捷地區分不同種類的樣品。關鍵詞：模化；單獨模型和聯合模型；樣品分類；二元判別分析法；模式識別；化學計量學Introduction0 Chemometrics is a branch of chemistry that uses statistical and mathematical methods for data analysis of chemical measurements[1~4]. It involves generation of powerful analytical methods and techniques for pattern recognition[5]. Compared to other chemometric pattern recognition techniques, classification is more useful under many circumstances as it can be applied to study the relationship between samples and their related classes rather than the relationship between different samples. In order to correctly classify unknown samples, appropriate models are assessed prior to classification. Different modelling approaches are required for classification of different samples and the best approach is determined on the basis of the merits of the modelling methods. Modelling approaches vary with different characteristics of the samples. In terms of the class (or classes) to which a sample is assigned, models are composed of hard models and soft models[6]. In terms of the class (or classes) that a model contains, models can be separated into disjoint and conjoint models[7~9]. In order to reduce the dimensionality of models, Principal Component Analysis (PCA) method is applied to construct PC models.[10~13] In contrast to visualisation of the models using the first two or three PCs, the optimal number of PCs for classification models is decided by the ratio of Predicted Residual Error Sum of Squares (PRESS) to Residual Sum of Squares (RSS) [5,14~15] us ing 200 boots t rap repetitions[16~18].Dis joint model l ing involves developing independent PC models for each group, and calculating both a separate D statistic[8, 19~23] about the classification performance of samples within the projected PC space and a Q statistic (square prediction error)[8,19~23] for each sample to the class model. The joint decision function obtained by the combination of D and Q statistics which is regarded as disjoint classifier is used to determine the assignment of samples. In contrast to disjoint model, conjoint model performs PCA on the entire analytical data to study the relationship between the samples and their class labels. Samples can be grouped into one of known classes according to the conjoint Quadratic Discriminant Analysis (QDA) classifier[5, 24~25]. In order to estimate the applicability and stability of disjoint and conjoint models, samples are randomly
18split into training set and test set[5, 25]. The training set is used to construct a model and test set is used to test the model. The iterative splits are randomly repeated 100 times and the overall model performance is associated with the average performance of 100 models. In order to prepare the information for classification, different data preprocessing methods[5] are applied to different data matrices. The applicability and stability of disjoint and conjoint models are examined by the performance indices of %PA (predictive ability) and %CR (classification rate)[5]. In this paper, pharmaceutical tablet samples analysed by liquid chromatography-mass spectrometry (LC-MS) were used to illustrate different modelling methods. In addition, the results of polymeric samples[7,26~32] obtained by Dynamic Mechanical Analysis (DMA)[26~28] were applied for comparison. The study of polymeric samples was to detect the separation of different groups so as to identify unknown polymeric samples for industrial application, whereas the separation of different classes is studied in order to assess the differences between each class for discrimination of tablets from different origins. Materials and Methods1 Materials and Software111 In order to demonstrate the modelling approaches for classification, 75 pharmaceutical tablet samples sourced from 3 different manufacturers (some tablets were manufactured from GSK company (G) and others were collected from companies in India (I) and Canada (C) were obtained using LC-MS profiles. The classification of different samples from different manufacturers was used to estimate the differences between different origins although all the samples were analysed at three different time points. In order to distinguish between samples of the same substances, the recognition and identification of impurities in tablet samples were performed by removing the main components and excipients. The Total Ion Currents (TICs)[33] of the remaining chromatograms were interpolated in order to form an analytical data matrix. The pharmaceutical datasets were collected from GlaxoSmithKline (GSK, Stevenage, Hertfordshire, UK) and all the experimental details had been discussed in previous literature[34]. Furthermore, polymeric samples which were acquired from Triton Technology Ltd (Keyworth, Nottinghamshire, UK) were also applied for comparison. All of the data analysis programmes were written in-house and applied using Matlab 7.6. (R2008a; MathWorks Inc, Natick, Massachusetts, USA)Modelling 111 Mode l l i ng was u sed a s a t e chn ique o f classification for delineating the relationship between the measurements and corresponding class membership. Since classification performance depended on model performance, different modelling approaches were adopted for determining the feasible and stable models for different classifications. When classification of samples from more than two classes is studied, disjoint and conjoint hard modelling methods were employed as it was quite difficult to assess models using ambiguous assignment. Because of the requirement of classification method, PCA was used to extract uncorrelated latent variables to construct PC models for data reduction so that the number of variables was fewer than the number of samples. Prior to the construction of PC models, the optimal number of principal components for each model should be determined by using the ratio of PRESS to RSS and 200 bootstrap repetitions. The constructions of disjoint and conjoint PC models were quite different. For disjoint PC models, PCA was performed on samples from one group or class and several separate ‘in group’ models were formed for disjoint classifiers, so a classification rule (decision function) was built according to the combination of all disjoint PC models. For conjoint model, PCA was performed on samples from all groups or classes and a single model was constructed for conjoint classifier.Classifiers113 For disjoint classifiers, disjoint models for all classes were scaled separately and they were compared in order to group samples into one of the classes for multi-class classification. The assignment of samples
19Dong LI,et al. Application of Disjoint and Conjoint Hard Models for Classification of Different Samples using Chemometric Pattern Recognitiondepended on the decision function constructed by the combination of the D statistic (QDA) and Q statistic. In contrast to SIMCA[8], disjoint classifiers were considered in a hard model framework and multiple one-class classifiers were applied. Disjoint QDA (one-class QDA) was used to group samples using the Mahalonobis distance to the centroid of each class. The distance of samples to each class was named D statistic. The Q statistic used the residuals between the PC model and its ‘in group’ samples. The class membership index of a sample was obtained according to the decision function formed by the combination of D and Q statistics. In order to determine the fitness of classified samples to each ‘in group’ model, the measures for the D and Q statistics as a ratio to their 95% confidence limit were combined together so that all classes were compared on a similar scale. The class membership of a sample was determined and the sample was assigned to the group or class with the minimum value of the decision function. For conjoint QDA, the Mahalonobis distance of analytical samples to the centroid of all classes were calculated and computed. Because a PC model was performed on all samples at once, only one PC model was available and Q statistic was not applicable. The class membership of a sample was determined by the conjoint QDA classifier and the sample was assigned to the group or class with the smallest distance of the sample to the centroid of the class. Classification114 Training and test sets114111 For determination of model performance, a portion of samples (training set) were used to establish a model and the remaining samples (test set) left out from the training set were used to assess the model performance by classifying samples excluded from the model. Apart from visualising the models by the first two or three PCs, the classification models might be composed of more than three PCs for each training set. When the number of PCs determined by bootstrap was greater than three, the model performance was reported in the form of various statistics following its elucidation by numerical indicators. Because the division of training and test sets were quite different for disjoint and conjoint classifications, different division procedures were demonstrated. For disjoint classification, approximately 2 / 3 of the samples from a specific class were used to establish a training set for the ‘in group’ class so that multiple training sets were constructed for all classes. A single test set was formed by using the remaining samples that were not selected for any of the training sets, so the same test set was applied for all training sets. Each model was used to predict the test set samples and the classification of a sample was determined by the combination of all modelling predictions. In contrast to disjoint classification, the number of samples from the smallest class was determined and approximately 2 / 3 of the number of samples from each class were assigned to the training set and the others were regarded as test set, so all the classes were represented equally in the training set and the conjoint model was balanced across all classes. For polymeric samples, the number of samples chosen for the training and test sets for both disjoint and conjoint classifications was demonstrated in previous literature[7]. For tablet samples, the number of samples chosen for the training and test sets for both disjoint and conjoint classifications is summarised in Table 1.Table 1. Number of samples in training and test set for tablet samples.Model Number of training set samplesNumber of test set samples (predictive modelling)Conjoint 33 42DisjointClass 1 G 25 25Class 2 I 11 (all models)Class 3 C 14For disjoint and conjoint classifications, the division of training and test sets was repeated 100 times and different samples were randomly selected for the training set. The overall classification performance was assessed over all 100 iterations. For both disjoint and conjoint classifications, 100 iterations were sufficient as all samples were selected for the training set many times.
20Preprocessing114111 In order to prepare the information and reduce the influence of factors such as large components (or variables), different data preprocessing methods were applied prior to classification of different samples. For the polymeric dataset, the vertical scale of the polymeric data had a physically interpretable meaning and the intensity was not influenced by the concentration of analytes or other factors, so mean centring[5] was applied to obtain a preprocessed data matrix as the variation around the mean value at each temperature point was characteristic of the polymers. [7] For the tablet dataset, the raw data matrix of tablet samples was composed of the TICs of impurities, so square root scaling[5] and row scaling[5] were performed on the raw data matrix. The final data preprocessing method was standardisation[5] and it was performed to form different preprocessed data matrices for disjoint and conjoint classifications. For disjoint classification, the training set for each ‘in group’ class was standardised and the test set was adjusted using the mean and standard deviation from the training set separately for different disjoint classifications. For conjoint classification, the mean and standard deviation calculated from the training set were used to standardise both training and test sets. Performance Indices111 The Percent Predictive Ability (%PA) and the Percent Classification Rate (%CR) were used to estimate the quality of classified results so that the applicability and stability of the disjoint and conjoint models could be assessed.[5, 7] Results 1 PC projections111 The distributions of all tablet samples in disjoint and conjoint PC space are illustrated using score plots of the first two PCs (Figure 1 and 2).For the tablet samples, the disjoint PC score plots (Figure 1) indicated that the ‘in group’ samples in each class could partly be distinguished from the ‘out group’ samples as some ‘in group’ samples overlapped with other samples. In general, samples from each ‘in group’ class could roughly be separated from other samples even though some misclassification might occur. ● Class1 (G) ■ Class2 (I) ◆ Class3 (C)Figure 1. PCA score plots of three disjoint models using the first two PCs for tablet samples (‘in group’ samples: close symbols; ‘out group’ samples: open symbols) from three different classes (G: blue; I: red; C: green).
21Dong LI,et al. Application of Disjoint and Conjoint Hard Models for Classification of Different Samples using Chemometric Pattern RecognitionAs far as the conjoint PC score plots for tablet samples (Figure 2) were concerned, all samples were present in the space of the first two PCs but most of them distributed close together. Although some samples were close together or even overlapped with each other, the separation between different classes could roughly be observed.● Class1 (G) ■ Class2 (I) ◆ Class3 (C)Figure 2. PCA plot of conjoint model using the first two PCs for tablet samples from three different classes.Classifications111 In contrast to the visualisation of disjoint and conjoint models using the score plots and the first two PCs, the number of PCs was optimised to construct PC models using bootstraps prior to disjoint and conjoint classifications. For the tablet samples, the optimal numbers of PCs for disjoint and conjoint models were quite similar and the most frequently chosen optimal number of PCs for disjoint and conjoint models for tablet samples is listed in Table 2.Classes Disjoint classification Conjoint classification1 G 22 I 1 13 C 1Table 2. Most frequently selected optimal number of PCs for disjoint and conjoint classifiers using 200 bootstrap iterations for tablet samples.In order to estimate the performance of disjoint and conjoint models, performance indices of the predictive ability and the classification rate were demonstrated for the test set samples. According to the histograms of disjoint and conjoint models (Figure 3), it was observed that disjoint models had better performance than conjoint models. Some samples from tablets I and C were completely misclassified (1) Disjoint QDA(2) Conjoint QDAFigure 3. Histograms of %PA for the test sets using disjoint and conjoint classifiers in the classifications of tablet samples from three different classes (G: blue; I: red; C: green).
22in the conjoint classification as they overlapped with other samples. According to the classified results of %CR for the test set samples in the classification of tablet samples, disjoint models had higher %CR for tablets G and C than conjoint models (Figure 4), but the %CR for tablet I in disjoint classification was slightly lower as more samples overlapped with others in disjoint models (Figure 1 and Figure 2). As shown in the histograms of %CR for both disjoint and conjoint classifications, most tablet samples could be grouped to their own classes. Figure 4. Histograms of %CR for the test sets using disjoint and conjoint classifiers in the classifications of tablet samples from three different classes (G: blue; I: red; C: green).
23Dong LI,et al. Application of Disjoint and Conjoint Hard Models for Classification of Different Samples using Chemometric Pattern RecognitionDiscussions3 PC projections311 For the polymeric samples, the score plots in disjoint PC space showed that the ‘in group’ samples could be distinguished from the ‘out group’ samples, but some groups had subgroups relating to the different grades of polymers. Although disjoint PC projections was used to demonstrate the structure of each polymeric group and the PC space was applied to best illustrate the ‘in group’ structure, some samples were assigned to more than one class according to the PC projections so that it might result in misclassification regardless of the variation between the PC models and the ‘in group’ samples. In terms of the conjoint score plots, the PC space was constructed for all polymeric groups using the first two PCs, but some samples overlapped with each other and this might also results in misclassification. Because all samples were represented in the same PC space, it was impossible to overcome the misclassification using Q statistic compared to disjoint PC models. For both disjoint and conjoint classification, better PC models could be established using greater number of PCs and bootstraps were applied to determine the optimal number of PCs for classification models.[7]Al though the overlapping might lead to misclassification of disjoint models for polymeric and tablet samples, Q statistic could be applied to distinguish the overlapping ‘in group’ samples for better classification of tablet samples as the overlapping ‘in group’ samples had smaller prediction error to the disjoint model than ‘out group’ samples. In the case of conjoint models for polymeric and tablet samples, Q statistic could not overcome the misclassification as all samples distributed in the same space. More PCs might be required to form better classification models for disjoint and conjoint classifications, and the optimal number would be decided by using bootstraps. Classifications311 For the polymeric samples, the optimal number of PCs for each disjoint model approximately corresponded to the grades of each ‘in group’ class. Compared to disjoint models, higher number of PCs was required to construct the conjoint model so that all samples could be adequately present in the same PC space[7]. In contrast to polymeric samples, the optimal numbers of PCs for disjoint and conjoint models for tablet samples were quite similar and they were not more than 2 because tablet samples did not have grades or subgroups in any class.Analogous to the histograms of %PA present in the classification of test set samples of tablets, disjoint models had better performance than conjoint models in the polymeric samples. Similar to the classified results of %CR for the test set samples in the classification of tablet samples, disjoint models had higher %CR for ABS and LDPE (groups 2 and 4) than conjoint models in the polymeric samples.[7] Compared to conjoint classification, the combination of D and Q statistics rather than a single statistic was applied for disjoint classification. Due to the incorporation of Q statistic into the disjoint models, test set samples could be classified as a class for disjoint classification according to the closest distance to the centroid of the class with smallest variation to the ‘in group’ model space. In contrast to disjoint classification, the distances of test set samples to each class was applied for classification but the variability between different classes was neglected. Thus, the %PA and %CR of disjoint models were better than conjoint models. Compared to conjoint models, disjoint models were more flexible for modelling the feature of each class and it was more stable as each disjoint model was not influenced by any ‘out group’ samples. However, conjoint classification was simpler and faster as a single PC model was performed for all classes.Concluding Remarks 4 Modelling plays an important role in classification for chemometric pattern recognition and the classification performance is associated with the model performance. Therefore stable and feasible modelling approaches are required for classification of different samples for different scientific purposes. In order to
24construct stable and feasible models for classification of multi-class samples, disjoint and conjoint modelling approaches were assessed. Comparing disjoint model with conjoint model, both modelling methods have advantages and disadvantages. For disjoint models, each disjoint model characterises the structure of an individual class, so the combination of disjoint models is more stable and flexible than a single model for overall classes as the ‘out group’ samples do not influence any ‘in group’ model and the misclassified samples which overlap with other samples may be distinguished by observing their variation to the ‘in group’ model. Moreover, disjoint model allows a new class to be modelled without recalculating the original models. For conjoint models, all classes are modelled simultaneously and all samples are present in the same space regardless of the variability of different classes, so it is much simpler and time-saving for the classification of multi-class samples. All results of disjoint and conjoint classifications indicated that the separation between different polymeric groups and tablet origins could be observed even though disjoint models had superior performance than conjoint models. For polymeric samples, disjoint models were more applicable and stable because the differences between different groups could be better extracted for the prediction of unknown polymeric samples in industrial applications. However, disjoint models were not quite feasible for tablet samples although the disjoint models are more stable with better classified results. In this case, the differences between different tablet origins could be clearly visualised by both modelling approaches, but disjoint models were more complex and time-consuming. Although better models are constructed for the classification of unknown tablet samples using disjoint models, conjoint models could simply and quickly be applied to observe the differences between different tablet origins. Disjoint modelling has advantage of constructing more stable and flexible models for classification and prediction of multi-class samples, whereas conjoint models can be simply and fast applied to recognise the differences between different classes. ReferencesD. L. Massart, B. G. M. Vandeginste, L. M. C. Buydens, S. De Jong, [1] P. J. Lewi, J. Smeyers-Verbeke, Handbook of Chemometrics and Qualimetrics Part A, Elsevier, Amsterdam, 1997.S. Wold, Chemometrics; what do we mean with it, and what do we [2] want from it? Chemom. Intell. Lab. Syst., 30, 109-115 (1995).B. R. Kowalski, Chemometrics, Anal. Chem., 52, 112-122 (1990).[3] R. Edmund, Malinowski, Factor Analysis in Chemistry (Third [4] Edition), Wiley, New York, 2002, Page 1.R. G. Brereton, Chemometrics for Pattern Recognition, Wiley : [5] Chichester, 2009.A. L. Pomerantsev, O. Y. Rodionova, Hard and soft methods for [6] prediction of antioxidants’ activity based on the DSC measurements, Chemom. Intell. Lab. Syst., 79, 73-83 (2005).D. Li, G. R. Lloyd, J. C. Duncan, R. G. Brereton, Disjoint hard [7] models for classification, J Chemometrics, 24, 273-287 (2010).S. Wold, Pattern-Recognition by Means of Disjoint Principal [8] Components Models, Pattern Recognition, 8, 127-139 (1976).M. P. Derde, D. L. Massart, UNEQ: A Disjoint Modelling Technique [9] for Pattern Recognition based on Normal Distribution, Anal. Chim. Acta, 184, 33 - 51 (1986).K. V. Mardia, J. T. Kent, J. Bibby, Multivariate Analysis, Academic [10] Press: London, 1979.S. Wold, K. Esbensen, P. Geladi, Principal Component Analysis, [11] Chemom. Intell. Lab. Syst., 2, 37-52 (1987).I. T. Jolliffe, Principal Components Analysis, Second Edition, [12] Springer: Berlin, 2002.B. F. J. Manly, Multivariate Statistical Methods: A Primer, Third [13] Edition, Chapman & Hall/CRC: New York, 2004.R. G. Brereton, Chemometrics: Data Analysis for the Laboratory and [14] Chemical Plant; Wiley: Chichester, 2003.S. Wold, Cross-Validatory Estimation of the number of components [15] in Factor and Principal Components Models, Technometrics, 20, 397-405 (1978).B. Efron, R. J. Tibshirani, An Introduction to the Bootstrap, Chapman [16] and Hall, New York, 1993.J. S. U. Hjorth, Computer Intensive Statistical Methods: Validation, [17] Model Selection and Bootstrap Chapman and Hall, New York, 1994.R. Wehrens, H. Putter, H. L. M. C. Buydens, The Bootstrap: a [18] Tutorial, Chemom. Intell. Lab. Syst., 54, 35-52 (2000).J. E. Jackson, A User’s Guide to Principal Components, Wiley: [19] Chichester, 2003.J. E. Jackson, Mudholkar GS Control Procedures for Residuals [20] Associated with Principal Components Analysis, Technometrics, 21, 341-349 (1979).P. Nomikos, J. F. MacGregor, Multivariate SPC Charts for Monitoring [21] Batch Processes, Technometrics, 37, 41-59 (1995).T. Kourti, J. F. MacGregor, Process analysis, monitoring and [22] diagnosis, using multivariate projection methods, Chemom. Intell. Lab. Syst., 28, 3-21 (1995).S. J. Qin, Statistical process monitoring : basics and beyond, J [23]
25Dong LI,et al. Application of Disjoint and Conjoint Hard Models for Classification of Different Samples using Chemometric Pattern RecognitionChemometrics, 17, 480-502 (2003).I. E. Frank, J. H. Friedman, Classification: Oldtimers and Newcomers, [24] J Chemometrics, 3, 463 - 475 (1989).S. J. Dixon, R. G. Brereton, Comparison of performance of five [25] common classifiers represented as boundary methods: Euclidean Distance to Centroids, Linear Discriminant Analysis, Quadratic Discriminant Analysis, Learning Vector Quantization and Support Vector Machines, as dependent on data structure, Chemom. Intell. Lab. Syst., 95, 1-17 (2009).B. E. Read, G. D. Dean and J. C. Duncan, Determination of Dynamic [26] Moduli and Loss Factors, Wiley: New York, 1991.J. C. Duncan, in Mechanical Properties and Testing of Polymers, [27] (Edited G M Swallowe), Kluwer Academic Publishers:Dordrecht, 1999.R. Faria, J. C. Duncan, R. G. Brereton, Dynamic Mechanical Analysis [28] and Chemometrics for Polymer Identification, Polym. Test. 26, 402-412 (2007).B. M. Lukasiak, S. Zomer, R. G. Brereton, R. Faria, J. C. Duncan, [29] Pattern Recognition for the Analysis of Polymeric Materials, Analyst, 131, 73-80 (2006).B. M. Lukasiak, S. Zomer, R. G. Brereton, R. Faria, J. C. Duncan, [30] Pattern Recognition and Feature Selection for the Discrimination between Grades of Commercial Plastics, Chemom. Intell. Lab. Syst. 87, 18-25 (2007).G. R. Lloyd, R. Faria, R. G. Brereton, J. C. Duncan, Learning [31] vector quantization for multi-class classification: application to characterization of plastics, J. Chem. Inf. Model, 47, 1553 -1563 (2007).G. R. Lloyd, R. G. Brereton, J. C. Duncan, Self Organising Maps for [32] distinguishing polymer groups using thermal response curves obtained by dynamic mechanical analysis, Analyst, 133, 1046-1059 (2008).W. Windig, W. F. Smith and W. F. Nichols, Fast interpretation of [33] complex LC/MS data using chemometrics, Anal. Chim. Acta, 446, 465-474 (2001).S. Zomer, R. G. Brereton, J. C. Wolff, C. Y. Airiau, C. Smallwood, [34] Component Detection Weighted Index of Analogy: Similarity Recognition on Liquid Chromatographic Mass Spectral Data for the Characterisation of Route/Process Specific Impurities in Pharmaceutical Tablets, Analytical Chemistry, 77, 1607-1621 (2005).教學/學術動態法學院2010/2011學年第二學期教學/學術活動一覽表序號 教學/ 學術活動 場次 日期 主題 主講者1 十字門法學論壇 第四場 1月18日 私有房屋公法限制相關法律問題2 十字門名家講壇 第五場 2月18日 中國土地改革的法學思考中國政法大學民商經濟法學院院長王衛國教授3 十字門法學論壇 第五場 3月14日 法律文化衝突現象與解釋4 十字門法學論壇 第六場 4月6日 刑事司法中的自由裁量5 學術講座 5月9日 內地刑事訴訟法修改的若干問題內地著名刑事訴訟法專案、中國政法大學陳光中教授6 十字門法學論壇 第七場 5月12日 公司控制權爭奪問題與法律對策 人文藝術學院2010/2011學年第二學期教學/學術活動一覽表序號 教學/ 學術活動 場次 日期 主題 主講者1 澳門文化產業發展高層論壇 第一屆 1月13日促進論壇成為一個政府、企業及專家學者共商澳門未來文化產業發展之對話平臺2 港澳傳媒名家講壇 第一場 1月24日 傳媒與生活品質香港中文大學社會科學院院長李少南博士3 港澳傳媒名家講壇 第二場 2月15日大都市、小地方：上海媒介生態的統合主義香港城市大學媒體與傳播學系主任李金銓博士4 人文藝術系列講座 第一場 3月22日 中國戲曲的藝術特徵國家一級演員、著名表演藝術家馬蘭教授5 人文藝術系列講座 第二場 3月22日 現代藝術的關鍵問題澳科大人文藝術學院院長、著名文化學者余秋雨教授6 人文藝術系列講座 第三場 3月23日 文化藝術中的風景園林中國工程院院士、風景園林規劃與設計教育家、北京林業大學孟兆禎教授7 澳門文化產業發展高層論壇 第二屆 5月19日討論澳門支柱產業 — 博彩業與文化產業的關係及發展前景
26第 5 卷 第 1 期 澳 門 科 技 大 學 學 報 Vol.5 No.12011 年 6 月 30 日 Journal of Macau University of Science and Technology June 30 , 2011 收稿日期：2010-11-18；修訂日期：2010-12-8作者簡介：劉超（1978—），男，澳門科技大學管理科學博士生，主要從事人力資源管理研究 E-mail：18998720119@189.cn，通訊地址：澳門氹仔偉龍馬路澳門科技大學行政與管理學院A408室領導風格、動態能力、創新導向文化 及組織績效關係劉 超（澳門科技大學行政與管理學院，澳門）摘要：首先提出領導風格對組織績效影響機制理論假設模型，並把組織學習能力、組織創新能力作為中介變數進行檢驗。然後該研究採用了描述性統計、信度效度分析、因子分析、多元回歸以及結構方程建模等統計技術，對澳門31家企業樣本資料進行實證分析。資料分析結果表明：領導風格通過組織學習能力、組織創新能力影響組織績效的理論假設模型成立，即任務導向型、關係導向型以及變革導向型領導風格會以組織學習能力為中介變數，進而對組織績效具有顯著影響；關係導向型、變革導向型領導風格以組織創新能力為中介變數，進而對組織績效產生顯著影響。任務導向型領導風格對組織創新能力沒有顯著影響。組織學習能力對組織創新能力具有顯著正向影響。創新導向文化對變革導向型領導風格與組織績效的關係具有顯著的調節效應。文章最後探討了本研究的貢獻和局限，並對澳門酒店企業領導風格改善提出建議。關鍵詞：領導風格；組織學習能力；組織創新能力；創新導向文化；組織績效Analysis on the Effects of Leadership Behavior Organizational PerformanceChao LIU( Macau University of Science and Technology, Faculty of Management and Administration, Macao, China )Abstract: This research presents a conceptual framework for organizational learning capability and organizational innovations capability as the mediating variable between leadership behavior and organizational performance. Based on the empirical studies and survey methodology, the sample of this study includes 31 enterprises in Macau Region to examine the interactive effect among organizational innovation capability, organizational learning capability, leadership behavior and organizational performance. Many quantitative methods, including Descriptive Statistics, Reliability Analysis, Factor Analysis, Multiple Regression and Structure Equation Model are used to test the hypothesis of the research. The major finding of this study is as following: Overall, the global model fit is acceptable. This empirical result supports the relationships among the variables. The three kinds of leadership behavior of the model have significant effect on the organizational learning capability, The two kinds of the change-oriented leadership behavior and the relation-oriented leadership behavior have significant effect on the organizational innovations capability, the task-oriented leadership behavior have not significant effect on the organizational innovations capability. In which the most important is change-oriented, the second is relation-oriented, and the third is task-oriented according to the significant. Data analysis showed that the moderating effects of innovation-oriented culture between leadership style and organizational performance. Organizational learning capability and organizational innovations capability have a partial mediate effect
27劉超 領導風格、動態能力、創新導向文化及組織績效關係引言0 澳門人多地狹，是世界著名旅遊都市，中國以及世界著名旅遊集團大多在澳門設點佈局，這裏具有完整的酒店形態與各種級別的酒店，該行業競爭特別激烈。因此，該區域企業樣本分佈廣泛，能夠較好地代表世界酒店業整體面貌。也正因為競爭異常激烈，該區域酒店服務業企業從高級管理層到基層普通員工很可能比中國任何其他區域的從業人員更需要高效的組織學習、組織創新以適應高度競爭的環境。組織領導人對組織具有重要的影響和獨特的貢獻，他們在組織資源配置、人才選拔培養和生產經營管理等方面被賦予最高拍板權，他們在組織權力分配中處於主導地位(Hambrick & Mason, 1984)[1]。當企業處在創業前期或業務轉型期，企業的制度和文化建設往往不盡完善，這使企業的發展很大程度上取決於企業家的領導水準。因此，研究其領導風格如何通過作用於組織學習能力、組織創新能力進而對組織績效產生影響，對該區域企業實際管理以及對中國更多企業的科學管理具有重要的指導作用。文獻探討1 學者們普遍認為領導風格與組織績效之間的關係不是直接的，而是通過影響企業的競爭性活動形成了一條複雜的資產和能力的鏈條，最終為企業帶來持續的競爭優勢。部分學者試圖從核心能力、組織學習行為、管理過程和競爭行動等視角，揭示領導風格對組織績效的影響機制，但是，這些研究並未揭示領導風格對組織績效的影響機制，很多研究仍處於探索性的階段，尚未得到普遍證實。在領導風格對組織績效影響的實證研究中有影響但不顯著、無影響以及有顯著影響的討論從來就沒有停止過(Slater & Narver,1994,1996; Pelham & Wilson,1996; Caruana & Berthon, 1996;Van Egeren & O'Connor,1998; Pitt & Berthon,1999)。有些學者試圖在上述關係中加入某些中介變數或調節變數來討論上述彼此矛盾的關係，例如組織創新行為 (Han, Kim & Srivastava,1998)、領導—下屬交換關係 (Shin & Zhou,2003; Ambalam Pushpanathan,2008)、組織學習活動(Versa & Crossan,2004)、組織承諾(Piccolo & Colquitt, 2006;)、組織公民行為（吳志明,武欣,2006）等[2~4]。但可以深入並能進行整體性分析領導風格、組織學習能力、組織創新能力以及組織績效四個方面關係的文獻，並能闡明領導風格對組織績效影響機制研究尚未出現。本文領導風格對組織績效影響機制1 研究變數界定111 領導風格研究普遍使用魅力型—任務型、交易型—變革型、專制型—關聯式的二維分類模型。二維領導模型雖然能夠深刻而有效地揭示領導風格的某些特點和本質，但由於領導風格的複雜性，將領導風格強行歸為非此即彼的二分法類別，可能產生歧義，並遺漏某些重要的領導風格，使定義的界定不準確(Yukl,1998,1999 ) [5~6]。本文借鑒Yukl (1998)對領導風格的定義，即按任務導向、關係導向和變革導向來劃分領導風格的三維領導風格模型。具體領導者的領導風格將涉及三種導向行為之一或是它們的一個組合。on the relationship between relation-oriented, change-oriented leadership behavior and organizational performance. Data analysis showed that organizational learning capability has positive direct significant impact on organizational innovation capability, but organizational innovation capability has not direct significant impact on organizational learning capability. Finally, the contribution and deficiency of the research are discussed. It provides managers with some valuable hints on how to improve organizational leadership behavior.Keywords: Leadership Behavior; Organizational Learning Capability; Organizational Innovation Capability; Innovation-Oriented Culture; Organizational Performance
28Richards（1997）在前人研究的基礎上，通過鑒別、測量促進組織學習的管理行為和重要特徵，提出了構成組織學習能力的五個維度，即目標任務共識、領導承諾與授權、試驗與鼓勵、知識傳遞能力和團隊工作能力。Goh（1998) [8]提出的學習型組織原型包括7個模組，其中5個核心的戰略模組是：願景與目標；參與和共同決策；組織文化；知識的轉化；團隊合作。兩個基礎模組是：可有效支援組織學習的組織結構設計；員工學習技能與素質。Goh根據這7個方面設計出一系列評價指標，以此評價某個組織的學習能力。陳國權（2005,2008）[9]認為組織學習能力體系由發現、發明、選擇、執行、推廣、反思、獲取知識、輸出知識和建立知識庫這9種相互影響的子能力構成。組織學習的總能力可以通過對這9種子能力加權平均計算出來。Watkins K. (1993) & Yang et al. (2004) 認為應該根據組織中個人、團隊以及組織三個層次分別設計學習能力測量指標並分別計量組織學習能力狀況。借鑒以上學者研究，本研究認為組織學習能力、組織創新能力是一種動態能力，是組織根據內外部環境變化，及時調用和部署組織學習資源、創新資源的能力。本文認為組織學習能力應該包括創建組織良好氣氛能力、學習成果分享能力、組織學習方向與內容的引導能力、組織與環境相協調能力(Watkins & Marsick, 1993, 1996, 1997; Marsick & Watkins, 1999, 2003; Yang et al. ,2004;戴萬穩, 2007)[10~13]。Daft(1978)認為組織創新從渠道來講可以分為組織內部經過組織學習自發產生和來自組織外部引入的創新；從內容來講可以包括管理創新及技術創新兩方面。本文認為組織創新包括組織從外部購進或由內部產生之現行的各項在規劃、組織、用人、領導、控制等企業活動方面的管理創新，以及在產品、制程及設備等的技術創新，而且必須已受組織成員肯定其貢獻度者（蔡啟通,1997）。組織績效指企業在單位時間創造的經濟效益和社會效益。為防止單指標測量組織績效可能具有的片面性，本文採用多指標來對其進行測量。本文採用Yang, Yu & Lee(2002);Baum & Wally(2003), Ravichandran & Lertwongsatien (2005)等學者對組織績效的測量，將組織績效分為知識管理績效、滿意度績效、經營績效[14]。圖1 本文研究模型表1 變革導向、關係導向及任務導向領導風格比較領導風格 任務導向型 關係導向型 變革導向型概念具有任務為核心，以績效為中心，重視賞罰重視業績考評，關心工作勝於關心員工與人際關係，監控任務導向表現的領導風格重視與下屬的互動以保持對組織目標任務的共識，重視領導承諾與授權，鼓勵下屬試驗、重視知識傳遞要素關係環境構建前攝的，喚起下屬產生卓越集體利益意識，幫助下屬達到非凡的目標，包含願景激勵、理想化影響力、感召力、智力激發、個性化關懷要素目標 超額任務、超額績效 領導下屬間良好關係品質在滿足成員的需求之外，更注重成員提升需求層次特徵從細節上規劃任務或專案，能準確地向下屬解釋任務或專案的責任及預期目標以明確下屬及領導各自扮演的角色及承擔的權利義務關係；經常根據計劃檢查工作進度，組織和協調例外工作，確保任務按計劃完成關心人際關係，建立並保持良性互動來營造最佳的關係氛圍；注重對下屬支持、信任、鼓勵，授權，善於營造比較寬鬆的組織氣氛。經常研究競爭對手的產品和行動以獲得改進組織的想法；總是提出新的和有創造力想法來改善產品、服務和流程；總是在提出變革的設想時表現得樂觀和充滿信心領域適用於維持組織的穩定與組織發展的正常運作適用於較為穩定且易於預測的環境適合於組織發生變遷或面臨危機時，為組織進行變革資料來源：本研究整理
36因此，組織學習能力與組織創新能力的展現要與企業文化相整合和適應。既定的企業文化也會影響人們對於領導風格模式的接受程度及其作用的發揮。第四，從資源觀理論角度分析，組織學習能力、組織創新能力都是某種動態能力，是組織根據內外部環境變化，及時調用和部署組織學習資源、創新資源的能力；組織學習能力及組織創新能力本身並不能產生持續的競爭優勢，上述能力的施展要充分考慮組織內外資源間的互補性和其他可能的調節變數，即要同組織內部其他的資源和能力相結合，包括企業文化、領導風格、環境因素等，當這種結合深深嵌入到企業中時，這種優勢會持續下去。之前文獻認為組織學習行為或組織創新行為是重要中介變數，並努力驗證其中介作用。但是本文認為組織學習行為和組織創新行為僅僅是組織的外在行為，由行為上升到組織能力以及到形成了核心競爭能力尚有很長的路要走。行為是多變的、易消逝的，特別容易受到外界干擾，但是能力是比較穩固的，一旦形成則不容易喪失，因此本文引入組織學習能力、組織創新能力作為重要中介變數。此外，雖然上述結論是在對澳門31家酒店企業進行調查研究的基礎上得出的，但是上述結論對其他行業其他區域的企業同樣具有借鑒意義，即本文的研究結論具有一定的普適性。所不同的是，對於不同行業與不同發展階段的企業而言，領導風格對組織績效的影響程度可能有一定的差異。但是，隨著經濟全球化進程的加快，領導風格在企業的生產、管理過程中將扮演越來越重要的角色，企業對提升領導風格有效性的要求也將越來越高，領導風格對組織績效的影響程度也將越來越大。致 謝非常感謝兩位匿名審稿人的評閱以及提出的寶貴意見。當然，文責自負。參 考 文 獻Hambrick D., Mason P., Upper Echelons: The Organizations a [1] Reflection of Its Top Manager. Academy of Management Review, 1984, 9 (2): 193-206.Slater S. F., & Narver J. C., Market Orientation and the Learning [2] Organization, Journal of Marketing, 1995 (59)3, 63-74.Shin, S. J. & Zhou, J. Transformational Leadership, Conservation and [3] Creativity: Evidence from Korea, Academy of Management Journal, 2003, (46), 703-714.Piccolo, R. F. & Colquitt, J. A. Transformational Leadership and Job [4] Behaviors: the Mediating Role of Core Job Characteristics, Academy of Management Journal, 2006, (49)2, 327-340.Yukl, G. Leadership in Organizations (5Th Ed.).(Englewood Cliffs, [5] NJ Prentice Hall, 1998: 59-64.Yukl, G. An Evaluative Essay on Current Conceptions of Effective [6] Leadership, Journal of Work and Organizational Psychology, 1999 (8)1, 33-48.Goh S., & Richards G. Benchmarking the Learning Capability of [7] Organizations, European Management Journal, 1997, (15)5, 575-583.[8]Goh, S. C. toward a Learning Organization: The Strategic Building [8] Block, Sam Advanced Management Journal, 1998, (63)2, 15-22.陳國權.複雜變化環境下人的學習能力:概念、模型、測量及影[9] 響.中國管理科學, 2008,16 (1):147-157.Watkins K. V. & Marsick. Sculpting the Learning Organization: [10] Lessons in the Art and Science of Systemic. San Francisco: Jossey-Bass, 1993.Watkins K. V. &Marsick. Adult Educators and the Challenge of the [11] Learning Organization. Adult Learning, 1996, 7(4):18-20.[12]Yang J., Yu L, Lee C. The Hidden Value of Knowledge in New [12] Products, Asia Pacific Journal of Management, 2002 (19), 573-586.戴萬穩.跨文化組織能力研究.南京:南京大學出版社, 2007.121-127[13] Baum, J. R. & Wally, S. Strategic Decision Speed and Firm [14] Performance, Strategic Management Journal, 2003, (24)11, 1107-1129.Dunn, S.C., Seaker, R.F. & Waller, M. A. Latent Variables in Business [15] Logistics Research: Scale Development and Validation, Journal of Business Logistics, 1994 ,(15)2, 145-172.MacKinnon, D. P., Warsi, G, and Dwyer J. H. A Simulation Study of [16] Mediated Effect Measures, Multivariate Behavioral Research, 1995 (30)1, 41-62.Baker W. E. & Sinkula J. M. Learning Orientation, Market Orientation [17] and Innovation: Integrating and Extending Models of Organizational Performance, Journal of Market-Focused Management, 1999, (4)4, 295-308.溫忠麟,張 雷,侯傑泰.有中介的調節變數和有調節的中介變數.心[18] 理學報,2006, 38(3):448-452.黃芳銘.結構方程模式理論與應用.北京:五南圖書出版股份有限公[19] 司, 2002. 67-73.
39第 4 卷 第 2 期 澳 門 科 技 大 學 學 報 Vol.4 No.22010 年 12 月 30 日 Journal of Macau University of Science and Technology Dec 30 , 2010 收稿日期：2011-3-1；修訂日期：2011-3-31本文為澳門基金會資助下的課題《澳門博物館的觀眾感知服務品質研究》的部份研究成果，課題編號為：0152。課題負責人為唐娟，課題組主要成員有李璽、王盈娟、王亮、王心、劉雨竹、李威朗。*通訊作者：唐娟，女，博士，助理教授，研究方向為：旅遊市場行銷。後博物館時代的澳門博物館發展展望唐 娟*，劉雨竹（澳門科技大學國際旅遊學院，澳門）摘要：在澳門大力推動產業多元化、發展旅遊文化產業的浪潮中，博物館作為記錄城市的“名片＂扮演著重要角色。本文首先對後博物館來臨的時代背景進行了剖析，並進一步在文獻綜述以及田野調查的基礎上，剖析澳門本土博物館的特徵以及面臨的種種問題，包括解說系統單薄、路標指引模糊、文字解說不足、商業模式單一。最後結合後博物館的時代背景提出若干啟示和展望，包括了服務品質導向、大眾文化導向、大博物館導向以及互動參與導向。希望對澳門及其他地區博物館業的發展有所實質性的幫助。關鍵詞：後博物館時代；澳門博物館；發展Some Insights into the Development of Museums in Macau: a Perspective of Post-museum EraJuan TANG, Yuzhu LIU( Faculty of International Tourism, Macau University of Science and Technology, Macao, China )Abstract: Macau Government is trying to develop industrial diversification and committed to establishing a city image of unique culture. On the basis of this background Museums are playing a vital role to represent the essential culture for the local region. The current post-industrial era embraces many changes for museum. Adopting the methodology of field study and depth interview with some experts in this circle, this study analyzes the features of museums in Macau. Some problems are then explored. To keep updated with the changes brought by the Post-museum era and settle the negative problems, some insights into the development of Museums in Macau are put forward in the end including service quality orientation, mass culture orientation, integration orientation as well as participation orientation. Keywords: Post-museum Era; Museums in Macua; Development引言：“後博物館時代”的來臨0 隨著社會的進步，人類的生活變得更加複雜多樣。博物館作為記錄社會的載體，區域文化的“名片”，在一個城市的歷史文化傳承，區域內與區域同區域之間人與人的理解和溝通，本地區人口文化素質的提高以及城市的宣傳與展示方面起到了重要作用。伴隨著後工業時代的到來，博物館事業的發展也進入到後博物館時期。其轉變主要體現在以下四個方面：第一，由精英文化向大眾文化轉變。博物館的種類已不僅限於法國羅浮宮，英國大英博物館，紐約大都會博物館這類館藏著稀世
41唐娟，等 後博物館時代的澳門博物館發展展望傳，在非物質文化遺產的創作、保護、保存和創新方面發揮越來越重要的作用，並使封閉性的博物館不斷轉向開放性的博物館，為豐富館藏文化做出更多的貢獻。二是作為文化教育的場所：博物館的社會功能最初主要是收藏和研究，教育是後來派生的功能。1880年，英國博物館學者魯金斯發表了《博物館之功能》的論文，強調博物館應成為一般公眾受教育的場所。美國學者顧迪在《將來的博物館》和《博物館行政管理的原則》兩篇專論中，進一步強調博物館必須致力於革新教育，開展積極的活動，使之不僅成為專家學者從事研究的場所，而且要成為教育機構的補充設施，校外教學園地[4]。20世紀以來，博物館在教育方面的價值越來越得以顯現。隨著類型和數量的不斷增多，博物館以其濃厚的藝術氛圍，豐富的實物收藏，不僅成為學校教育生動的校外課堂，也是成人不斷求知的場所。三是公共交流的視窗：博物館從誕生之初，就始終承擔著傳播科學文化的任務。這種傳播和交流，既存在於地區民眾之間，也存在於各國、各地區和各民族之間。它以一種集中展現文化的形式，成為增進瞭解和友誼的視窗和紐帶。紐約現代藝術博物館的收藏在藝術收藏界享有盛譽，其中包括著名畫家塞尚、凡高、畢卡索、米羅和達利等的很多極為重要作品。去年，紐約現代藝術博物館由於內部整修，把主要的繪畫和雕塑展品都運到柏林進行展覽。在柏林新國家藝術畫廊展出的“MOMA在柏林”盛況空前，成為歐洲藝術界的一大盛事。展覽舉辦20天時，參觀人數即突破10萬，在第163個展出日，觀眾人數更是累計達到了100萬名[5]。在這些參觀者中，不僅有柏林市民，也有從德國各地慕名而來的藝術愛好者，更有來自世界各地的遊客。博物館把人們從各地聚集到一起，為人們提供了增進瞭解和學習交流的機會，促進了博物館休閒娛樂活動的產生和發展。博物館與旅遊的關係11111 近年來，博物館與旅遊的關係一直是博物館學界和旅遊界研究的課題。布魯納認為，博物館與旅遊有著類似之處1。Silberberg[6]研究了城市背景下博物館和遺產地文化旅遊及商機問題，認為文化旅遊給博物館和遺產地帶來重要的經濟利益，但博物館和遺產地文化旅遊需要有關政策和實踐保障，如吸引遊客、社區合作和市場行銷等，同時需要對經營者進行遺產保護和相關教育，提出文化旅遊產品的8 個評估要素2。此外，文章將文化旅遊市場概括為5 大圈層（見表1），提出博物館等文化產品增加市場份額的策略是將文化的吸引力從僅佔市場份額小部分的完全動機型旅遊者，擴展到更大份額的部分動機型旅遊者，附屬動機型旅遊者以及偶然旅遊者。表1 按動機程度劃分的博物館文化旅遊市場圈層 動機類型 具體表現第一圈層完 全 動 機 型 ： 文化因素強烈驅動到某城市旅遊很明確的是由於這個城市的劇場環境，博物館以及文化節等第二圈層部 分 動 機 型 ： 部分 動 機 為 文 化 因素到某地旅遊既是因為當地的文化環境和氛圍，也可能是出於探親訪友等其他目的第三圈層附 屬 動 機 型 ： 文化 動 機 僅 僅 是 另一 主 要 動 機 的 附屬旅遊者訪問該城市的主要動機可能是非文化的，但是旅遊者的旅行計劃中，也包含了文化環境第四圈層偶 然 動 機 型 ： 文化 動 機 是 偶 然的，臨時的文化旅遊的行為不是計畫當中的，是偶然發生。例如，並不打算參觀文化吸引物或事件，但朋友、親戚陪同前去或者恰好文化環境在賓館附近圈外無 動 機 型 ： 無 任何文化動機 任何情況下，都沒有參觀文化吸引物或者事件資料來源：Ted Silberberg. Cultural tourism and business opportuni-1. Bruner E ed. Special issue museum and tourism [J].Museum Anthoropolgy,1993a,17(3) 2 . Te d S i l b e r b e r g . C u l t u r a l t o u r i s m a n d b u s i n e s s opportunities for museums and heritage sites[J]. Tourism Management,1995,16(7):491—500
42ties for museums and heritage sites [J]. Tourism Management, 1995, 16(7):497Lee Jolliffe和Ronnie Smith[7]通過對北大西洋兩個島嶼的案例研究，具體探究了旅遊與博物館的關係，分析了各自在文化遺產中的地位（見表2）。表2 旅遊與博物館的關係檢測項目 屬性 內容檢測1 相似性兩者均由旅行產生；產生和展示文化；構建設和編錄文化；組織和產生體驗；依賴觀眾檢測2 差異性旅遊的目標一般是效益；博物館一般為非營利機構檢測3 交叉性旅遊者訪問博物館；博物館為旅遊者展示人類遺產檢測4 理念博物館採用不同的觀念；博物館採取不同的活動的認知檢測5 開發內容目的地範疇：管轄權的因素；經濟發展；資源支援資料來源：Lee Jolliffe, Ronnie Smith. Heritage, tourism and museum: the case of the North Atlantic islands of Skye, Scotland and Prince Ed-ward Island, Canada [J].International Journal of Heritage Studies, 2001, 7(2):154.博物館旅遊者行為的研究11113 Andrea Davies和 Richard Prentice[8]在1995年針對細分的博物館旅遊市場，在胡德（Hood）以訪問頻率對旅遊者進行區分的基礎上，將經常性旅遊者，偶然性旅遊者以及無遊覽行為的旅遊者這三類博物館觀眾細分群體中的無遊覽行為旅遊者（非訪問者）做了更進一步的擴展。通過問卷調查及資料分析， Myriam Jansen-Verbeke和Johan Van Rekom[9]於1996年研究了城市博物館旅遊者的旅遊動機，結果顯示旅遊者訪問鹿特丹美術館（Rotterdam Fine Arts Museum）的比例高達54%；並採用了梯級洞察方法（Laddering technique）分析收集的資料，證實了博物館作為城市旅遊產品的關鍵要素所起到的作用，並從旅遊者的動機角度提出了博物館旅遊的行銷戰略。關於旅遊者對博物館的期望， Julia Harrison[10]以夏威夷一博物館為例進行了實證研究，獲得一組有關博物館旅遊者的定性和定量的資料，研究發現，人們期望博物館能植根於當地的經濟和文化之中，反映出地方的特色文化，體現“當地性”以及當地居民參與博物館規劃與發展的積極意義。在旅遊體驗中，除有形產品以外，無形的情感體驗對旅遊者來說更為重要， Richard Prentice[11]從旅遊體驗的真實性出發，提出新的博物館行銷概念“同心圓模型”（從內到外依次為旅遊者體驗、無形構建、有形構建），闡述了博物館在無形產品的開發中如何體現真實性。國內博物館研究111 博物館與旅遊的關係11111 陸建松[12~13]曾多次對行業博物館進行專門研究，認為在保持博物館文化展示個性的前提下，充分發揮其旅遊觀光功能和商貿博覽功能，既可以保障行業博物館的經費來源，又是旅遊和商貿活動發展的需要。陳朝隆等學者[14]基於國際視角，提出了我國城市博物館旅遊開發策略，並論述了博物館在城市旅遊中的定位、博物館門票以及旅遊與博物館合作等若干現實問題。張敏[15]提到，博物館與旅遊業是兩種性質不同的行業，博物館是公益性文化事業，追求的是社會效益；旅遊是經營型企業，以追求利潤為目的。但是他們服務的物件是廣大群眾，這是一致的。博物館和旅遊業各自按照自己的特點、規律去運行，同時互相配合和支持，從而獲得雙贏的結果。博物館經營研究11111 烏曼菁[16]討論了博物館紀念品的開發等問題，並進一步從行銷學的角度分析了博物館商店的經營。林美珍、鄭向敏等[17]提出，博物館參與會展旅遊能夠獲得政策優勢、成本優勢、專業化優勢以及資源利用優勢，但受傳統經營體制的影響，博物館與市場嚴重脫節。隨後談到博物館經營理念分析，包括市場化運作、企業化經營、數位化管理、國際化拓展理念。提出了包括產品策略、行銷策略、合作策略在內的經營策略。孫旭紅，孫巨實[18]指出旅遊產品應包括三個層次的內容，即核心產品、有形產品和附加產
46互動參與導向314 數字科技和娛樂休閒的融合31411 越來越多的數字科技進駐博物館，加強了觀眾的參與性。在美國國家自然歷史博物館中，可以用iphone拍照出自己的照片傳輸給現場的機器，經過機器的處理便可看出自己在遠古時代的樣子，再如可以自己製造地震波等，都是現代科技的應用。在娛樂休閒方面，國外許多博物館中都設有兒童探索室，低齡兒童可以在工作人員的指導下做一些活動。在美國博物館參觀時，發現很多牌子上寫的是“請觸摸”，這與澳門博物館中常見的“請勿觸摸”截然相反，而與展品或樣本觸摸的過程本身就增加了與觀眾的互動性及娛樂性。將靜態的博物館動起來、從“請勿觸摸”向“請觸摸”轉變，是將是澳門博物館的發展方向之一。空間安排和路線的組合31411 博物館場館空間的安排和參觀路線的設計和組合也很重要。一個合理且富有吸引力的路線可以方便觀眾的參觀。在美國國家自然歷史博物館中，你可以很隨意的挑選路線走，一旦選擇方向，展覽物品會吸引你繼續往前走，而且不會走回頭路。出口及幾個館之間的連接非常緊密，每個館之間從大體上具有相關性，如博物館第一層的西側主要為近現代的動物和人類，而東側則主要是遠古時代的動物展出。但是，每兩個相鄰館的裝潢風格、顏色和內容又有很大的差異，給參觀者不斷地新鮮感，保持參觀者的參觀興趣。而在紐約大都會博物館，場館服務人員會主動向你提供幫助，他們會根據你的興趣和你打算參觀的時間來幫你設計參觀路線。 聲、光、味的結合31413 聲音、光線、氣味作為輔助性設施將其合理的運用可以突出展品和創造符合博物館風格的環境。博物館可以根據展品的主題配合相應音樂，宜人的香味，合適的燈光。在將你帶入間諜博物館的電梯裏，閃亮的霓虹燈有節奏的閃爍著，一下子把你帶入緊張、激動的氣氛中。而自然歷史博物館的非洲動物館裏，你能聽到各種動物的叫聲，生動刺激。每隔一段時間播出的雷雨聲音，將觀眾帶入了一片熱帶雨林的遐想中。澳門博物館也有很好例子，文化會館配有悠揚的音樂，怡人心脾；墓地中配有悲樂，符合展示主題；進入仁慈堂有一股淡淡的檀香味，與古老的建築風格和室內館藏相互呼應。因而，增添燈光、聲音和氣味的渲染使得展品更具有感染力。參 考 文 獻Awoniyi Stephen. The contemporary museum and leisure: recreation [1] as a museum function [J]. Museum Management and Curatorshi, 2001, 19(3): 297-308.Eilean Hooper-Greenhill. Museums and Their Visitors [M]. [2] London,Routledge.George F. MacDonald, Stephen Alsford. The museum as information [3] utility [J]. Museum Management and Curatorship, 1991, 10(3): Pages 305-311.博物館綜論 http://art.tnca.edu.tw/museum/.[4] 柏林官方中文網－新柏林http://www.berlin-china.net/news/index.[5] php?newsid=115) Bruner E ed. Special issue museum and tourism [J].Museum Anthoropolgy, 1993a, 17(3) Ted Silberberg. Cultural tourism and business opportunities for museums and heritage sites[J]. Tourism Management, 1995, 16(7):491—500.Ted Silberberg. Cultural tourism and business opportunities for [6] museums and heritage sites [J]. Tourism Management, 1995, 16(7):497.Lee Jolliffe, Ronnie Smith. Heritage, tourism and museum: the case [7] of the North Atlantic islands of Skye, Scotland and Prince Edward Island, Canada [J].International Journal of Heritage Studies, 2001, 7(2):154.Andrea Davies, Richard Prentice. Conceptualizing the latent visitor to [8] heritage attractions [J]. Tourism Management, 1995, 16(7): 491—500.Myriam Jansen-Verbeke, Johan Van Rekom. Scanning museum [9] visitors Urban tourism marketing [J]. Annals of Tour-ism esearch, 1996, 23(2):364—375.Julia Harrison. Museum and touristic expectations [J]. Annals of [10] Tourism esearch, 1997, 24(1):23—40.Richard Prentice. Experiential Cultural Tourism: Museums & the [11] marketing of the New Romanticism of evoked authenticity [J]. Museum Management and Curatorship, 2001, 19(1):5—26.陸建松.博物館與都市旅遊業[J].探索與爭鳴,1997(11):37-38.[12] 陸建松.上海市行業博物館建設意義、現狀及其存在問題思考[J].[13] 復旦學報(社會科學版)2003(4):130-136.陳朝隆, 陳烈等.城市博物館旅遊淺議[J].桂林旅遊高等專科學校學[14] 報,2006(3):361-365.張敏.博物館與旅遊[J].中國博物館，2004，（1）：24—28.[15] 烏 曼 菁 . 從 行 銷 學 角 度 淺 析 博 物 館 商 店 的 經 營 [ J ] . 博 物 館 研[16] 究,2005(1):73-76 .鄭向敏, 林美珍, 陳傳錢, 澳門"賭城"旅遊形象的SWOT分析[J].旅[17] 遊科學，2004,18（3）：1-7.
48第 5 卷 第 1 期 澳 門 科 技 大 學 學 報 Vol.5 No.12011 年 6 月 30 日 Journal of Macau University of Science and Technology June 30 , 2011 收稿日期：2011-3-1；修訂日期：2011-3-27*王心，女，現為澳門科技大學國際旅遊學院全日制博士研究生。曾任雲南大學編輯部部長、《東陸英才》報主編、《鼎元》雜誌社特約責編等職務。並於2008年取得雲南大學新聞專業文學學士學位、於2010年取得澳門科技大學國際旅遊管理碩士學位。目前的主要研究方向為：慶典管理和事件旅遊。並已就相關研究領域撰寫文章，分別在《中國會展》、《中國會議》、《商業研究》等主要刊物作發表。E-mail: xwang445@yahoo.com.hk TEL: （853）6218 0445 ；（86）153 4488 0445澳門美食節嘉年華顧客滿意度研究王 心（澳門科技大學國際旅遊學院，澳門）摘要：本文旨在研究澳門美食節顧客滿意度，探究影響其表現性的各種因素，找出澳門美食節今後努力的重點和方向。為達上述目的，本研究通過對國內外相關文獻進行綜述、田野調查的方式，針對澳門第九屆美食節嘉年華做實證研究。本研究通過問卷調查，測試顧客對澳門美食節各項因素的重要性和績效之間的全面差異，歸納影響美食節顧客滿意度的主要因素；并引入重要性績效分析（IPA），用以瞭解提升美食節全面顧客滿意度應該著重於哪些方面。並根據研究結論提出建議，以供相關部門、業者以及未來相關研究之參考。關鍵詞：美食節；顧客滿意度；澳門A Study on Customer Satisfaction of Macau Food CarnivalXin WANG( Faculty of International Tourism, Macau University of Science and Technology, Macao, China )Abstract: This paper aims at studying on customer satisfaction of Macau Food Festival and the various factors affecting its performance, and finding the key point and direction of Macau food festival. To achieve these ends, this study does empirical research on Macau 9th Food Carnival through doing overview of the relevant literature in and abroad and field research. It through the questionnaires to test the overall difference between importance and performance and sum up the main factor affecting the customer satisfaction of Food Festival. Then it introduces the IPA to understand which fields we should emphasize to improve the overall customer satisfaction of Food Festival. And according these study results, it states some suggestions for reference of relevant departments, practitioners and future relevant research.Keywords: Food festival; Customer satisfaction; Macau引言0 澳門自古以來是一個中西文化交匯的地方，其中包括飲食文化的交流。中西飲食文化互相交融，除了傳統的中國美點外，在澳門還有西式美食、東南亞美食、日本美食、以及結合中西飲食文化獨具本地特色的澳門式葡國大餐。澳門憑藉其獨特的美食資源，已被《米芝蓮指南》列為美食搜羅目的地，發行了《米芝蓮指南香港澳門2010》。米芝蓮星級的頭銜彷彿等同飲食界的至高榮譽，叫人趨之若鶩。不少遊客手持旅遊書、美食指南等慕名來澳，每逢節假日書上介紹的食肆幾乎都擠滿了遊客，氹仔市區的豬扒包便是代
57王心 澳門美食節嘉年華顧客滿意度研究更長的路、更多的挑戰，我將時刻以母校為驕傲，一如既往地積極向上、努力前行，爭取做出更大的成績。參 考 文 獻章采烈（1997）。中國美食特色旅遊，北京:對外經濟貿易大學[1] 出版社.邵萬寬（2000）。美食節策劃與運作，瀋陽：遼寧科學技術出版[2] 社. Kim, Y. (2007). An investigation of food tourism at a food event: [3] an examination of food tourists＇ behavior. Texas : The Texas Tech University.Wolf, E. (2002). Culinary Tourism: A Tasty Proposition, URL [4] (consulted 14 March 2005): www.culinarytourism.org/faq.php.Long, L. (2004). Culinary Tourism. Lexington, KY: The University [5] Press of Kentucky.Hall, C. M., & Mitchell, R. (2001). Wine and food tourism. In N. [6] Douglas & R. Derret (Eds.), Special interest tourism: Context and case:307-329. Brisbane: John Wiley & Sons.Hall, C. M., & Sharples, L. (2003). The consumption of experiences [7] or the experience of consumption? An introduction to the tourism of taste. In C. M. Hall, L. Sharples, R. Mitchell, N. Macionis & B. Cambourne (Eds.), Food tourism around the world: Development, management and markets:1-24. Oxford: Butterworth-Heinemann.du Rand, G., Heath, E., & Alberts, N. (2003). The role of local and [8] regional food in destination marketing: A South African situation analysis. Journal of Travel and Tourism Marketing, 14:97-112.Cela, A., Knowles-Lankford, J., & Lankford, S. (2007). Local food [9] festivals in Northeast Iowa communities: A visitor and economic impact study. Managing Leisure, 12:171-186.Quan, S., & Wang, N. (2004). Towards a structural model of the [10] tourist experience: an illustration from food experiences in tourism. Tourism Management, 25(3), 297-305.Hjalager, A., & Richards, G. (2002). Tourism and gastronomy. [11] London: Routlege.Martilla, J. & James, J. (1997). Importance-Performance Analysis. [12] Journal of Marketing, 41(1):77-9.Sylvia Smith & Carol Costello (2008). Culinary tourism: Satisfaction [13] with a culinary event utilizing importance-performance grid analysis. Journal of Vacation Marketing, 15(2), 99-110.Ticehurst, Gregory W. & Veal, Anthony J. (1999). Business Research [14] Methods: A Managerial Approach. Australia: Longman.第二屆校園詩歌朗誦大賽成功舉行由通識教育部主辦的“感恩之心”——第二屆校園詩歌朗誦大賽5月7日晚於科大D 座禮堂舉行。比賽設粵語組和普通話組，校內、外特聘專家擔任了評委。5月7日恰逢母親節前夕，這次以“感恩之心”為主題的校園詩歌朗誦大賽，為科大學子提供了一個感謝父母、感謝母校、感恩社會的契機。該賽事由通識教育部語文組於今年1 月份開始籌備，吸引了眾多在校大學生踴躍報名參加。今年3 月和4 月已分別舉行了初賽和半決賽，從百餘組參賽選手中篩選出了25組進入決賽。語文組老師還多次對選手進行了一對一的指導和培訓。校園詩歌朗誦大賽始於去年，為慶祝科大成立十周年而成功舉辦。在校領導的大力支持下，今年繼續舉行。本屆較上屆規模更大，師生反映熱烈。決賽現場座無虛席，整個比賽持續近兩小時。詩情的演繹、精美的畫面、動聽的音樂、優美的舞蹈交織成現場浪漫激情的氛圍。評委和嘉賓分別對是次比賽進行了精彩的點評，並嘉許科大是一個富有文化底蘊又充滿生機和活力的大學。最後許敖敖校長表示，科大校園文化建設取得了豐碩的成果，學生在國內外諸多比賽中屢創佳績。本次朗誦比賽作為科大校園的重要文化活動，充分展示了科大第二課堂的優秀成果。他感謝同學們的積極參與及通識教育部語文組老師們的辛勤勞動。本次朗誦大賽設置的獎項有：粵語組和普通話組一二三等和優勝獎，原創詩歌獎；此外還增加了由學生評審團選出的部分獎項。校園文化
58第 5 卷 第 1 期 澳 門 科 技 大 學 學 報 Vol.5 No.12011 年 6 月 30 日 Journal of Macau University of Science and Technology June 30 , 2011 收稿日期：2010-10-15；修訂日期：2011-4-15* 通訊作者：丁浩，男，管理學博士，澳門科技大學行政與管理學院助理教授，主要研究方向為金融風險管理、金融工程與投資管理。Email: hding@must.edu.mo基於非對稱GARCH-VaR模型的中國利率風險度量 ——對Shibor的實證丁 浩，龐 川（澳門科技大學行政與管理學院，澳門）摘要：利率風險是商業銀行面臨的最重要的市場風險，由於中國仍然對存貸款利率管制，選擇市場化程度最高的同業拆借市場研究利率風險。通過建立正態、t和GED分佈下反映信息非對稱性的TARCH、EGARCH和PARCH模型度量利率的條件波動率，進而建立條件VaR模型以度量因利率風險而導致的資產損失大小，最後基於Shibor數據對中國利率的總體風險情況進行了實證研究。關鍵詞：利率風險；VaR；GARCH；ShiborEstimating China's Interest Rate Risk Based on Asymmetric GARCH-VaR Models: An Empirical Analysis of ShiborHao DING, Chuan PANG( Faculty of Management and Administration, Macau University of Science and Technology, Macao, China )Abstract: Interest rate risk is the most significant market risk for commercial banks. As the deposit and lending rates are still controlled in China, we choose the interbank lending market which is highly marketized to study the interest rate risk. The GARCH family models, which reflect the asymmetric information under normal, t and GED distribution, are established to describe the conditional volatility of interest rate. Furthermore, the conditional Value-at-Risk model is built to measure the expected loss caused by interest rate risk. Finally, an empirical analysis is given with the data from Shibor system.Keywords: Interest Rate Risk; Value at Risk; GARCH; Shanghai Interbank Offered Rate引言0 利息是投資人讓渡資本使用權而索要的對機會成本和風險的補償，利率反映了這種補償的大小，是資本的價格。作為生產的三大要素之一，資本因其稀缺性而須得到有效配置，通過供求雙方力量對比形成均衡價格的市場方式，被證明是有效率的方式。然而利率管制導致了資本價格的扭曲，利率無法成為資本市場供求關係的傳遞信號。對利率管制的放鬆過程被稱為利率市場化，利率市場化的最終目的是實現利率由資本市場供求決定，通過有效地配置資本，進而運用到實體經濟中去。利率市場化機制在提高資本配置效率
67丁浩，等 基於非對稱GARCH-VaR模型的中國利率風險度量 ——對Shibor的實證致 謝作者感謝匿名評審人提出的中肯建議。當然，文責自負。參 考 文 獻Global Derivatives Study Group. Derivatives: Practices and Prin-[1] ciples. Washington D C, 1993.J. P. Morgan. RiskMetrics Technical Document. Morgan Guaranty [2] Trust Company, 1994.Jorion P. Value at Risk: The New Benchmark for Controlling Deriva-[3] tives Risk. New York: McGraw-Hill, 1997.Penza P and Bansal V K. Measuring Market Risk with Value at Risk. [4] New York: Wiley and Sons, 2001.Jorion P. Value at Risk: The New Benchmark for Managing Financial [5] Risk, 3rd ed. New York: McGraw-Hill, 2007.Krehbiel T and Adkins L C. Extreme Daily Changes in U.S. Dollar [6] London Inter-bank Offer Rates. International Review of Economics and Finance, 2008, 17: 397-411.Ferreira M A and Lopez J A. Evaluating Interest Rate Covariance [7] Models Within a Value-at-Risk Framework. Journal of Financial Econometrics, 2005, 3(1): 126-168.Vlaar P. Value at Risk Models for Dutch Bond Portfolios. Journal of [8] Banking and Finance, 2000, 24: 1131-1154.Lin C H, Chang Chien C C and Chen S W. A General Revised Histori-[9] cal Simulation Method for Portfolio Value-at-Risk. The Journal of Alternative Investments, 2005, 8(2):87-103.Fiori R and Iannotti S. Scenario-based Principal Component Value-[10] at-Risk When the Underlying Risk Factors are Skewed and Heavy-tailed: An Application to Italian Banks＇ Interest Rate Risk Exposure. Journal of Risk, 2007, 9(3): 63-99.Yule G U. On a Method of Investigating Periodicities in Disturbed [11] Series, with Special Reference to Wolfer＇s Sunspot Numbers. Philosophical Transactions of the Royal Society, 1927, 226: 267-298.Walker G. On Periodicity in Series of Related Terms. Proceedings of [12] the Royal Society, 1931, 131: 518-532.Box G E P and Jenkins G M. Time Series Analysis Forecasting and [13] Control. San Francisco: Holden-Day, 1970.Mandelbrot B B. The Variation of Certain Speculative Prices. The [14] Journal of Business, 1963, 36(4): 394-419.Fama E F. The Behavior of Stock Market Prices. The Journal of [15] Business, 1965, 38(1): 34-105.Engle R F. Autoregressive Conditional Heteroscedasticity with Es-[16] timates of the Variance of United Kingdom Inflation. Econometrica, 1982, 50(4): 987-1008.Bollerslev T. Generalized Autoregressive Conditional Heteroskedas-[17] ticity. Journal of Econometrics, 1986, 31: 307-327.Black F. Studies of Stock Market Volatility Changes. Proceedings of [18] the 1976 Meetings of the American Statistical Association, Business and Economics Statistics Section, 1976: 177-181.Zakoian J M. Threshold Heteroskedastic Models. Unpublished manu-[19] script, CREST, INSEE, 1990.Zakoian J M. Threshold Heteroskedastic Models. Journal of Econom-[20] ic Dynamics and Control, 1994, 18: 931-995.Glosten L R, Jagannathan R and Runkle D E. On the Relation between [21] the Expected Value and the Volatility of the Nominal Excess Return on Stocks. Journal of Finance, 1993, 48(5): 1779-1801.Nelson D B. Conditional Heteroscedasticity in Asset Returns: A New [22] Approach. Econometrica, 1991, 59(2): 347-370.Taylor S. Modeling Financial Time Series. New York: Wiley and [23] Sons, 1986.Schwert G M. Why does Stock Market Volatility Change over Time. [24] Journal of Finance, 1989, 44: 1115-1153.Ding Z X, Granger C W J and Engle R F. A Long Memory Property [25] of Stock Market Returns and a New Model. Journal of Empirical Finance, 1993, 1: 83-106.Brooks R, Faff R, Mckenzie M and Mitchell H. A Multi-Country [26] Study of Power ARCH Models and National Stock Market Returns. Journal of International Money and Finance, 2000, 19(3): 377-397.Giot P and Laurent S. Value-at-Risk for Long and Short Trading [27] Positions. Journal of Applied Econometrics, 2003, 18(6): 641-664.譚地軍, 田益祥, 黃文光. 有信息衝擊、無信息衝擊與波動率非對[28] 稱性. 管理工程學報, 2009, 23(2): 92-98.Johnson C and Soriano F. Volatilidad del Mercado Accionario y la [29] Crisis Asiática: Evidencia Internacional de Asimetrías. El Trimestre Económico, 2003, 71: 355-388.Engle R. Risk and Volatility: Econometric Models and Financial [30] Practice. American Economic Review, 2004, 94(3): 405-420.Jayasuriya S A. Asymmetric Volatility in Asian Equity Markets. Jour-[31] nal of International Finance and Economics, 2008, 8(4):11-21.Fan L Z and Zhang C. The Chinese Interbank Repo Market: An [32] Analysis of Term Premiums. The Journal of Futures Markets, 2006, 26(2): 153-167.Longstaff F. The Term Structure of very Short Rates: New Evidence [33] for the Expectations Hypothesis. Journal of Financial Economics, 2000, 58: 397-415.蔣競. 我國利率市場化中基准利率的選擇與培育. 西南交通大學學[34] 報(社會科學版), 2007, 8(5): 28-33.Imam M. The Chinese Interbank Markets: Cornerstone of Financial [35] Liberalization. China & World Economy, 2004, 12: 17-33.李志輝, 劉勝會. 我國商業銀行利率風險的度量研究. 南開經濟研[36] 究, 2006, (3): 27-41.李成, 馬國校. VaR模型在我國銀行同業拆借市場中的應用研究. 金[37] 融研究, 2007, (5): 62-77.Ding H and Xu C X. Measuring the Risk of China Interbank Lending [38] Market in VaR. Advances in Business Intelligence and Financial Engineering, Paris: Atlantis Press, 2008.Jorion P. Risk2: Measuring the Risk in Value at Risk. Financial [39] Analysts Journal, 1996, 52(6): 47-56.Engle R F and Ng V K. Time-Varying Volatility and the Dynamic Be-[40] havior of the Term Structure. Journal of Money, Credit and Banking, 1993, 25(3): 336-349.Zivot E and Wang J H. Modeling Financial Time Series with S-PLUS [41] (2nd ed.). New York: Springer, 2006.Shively P A. Asymmetric Temporary and Permanent Stock-price In-[42] novations. Journal of Empirical Research, 2007, 14 (1): 120-130.Kupiec P. Techniques for Verifying the Accuracy of Risk Measure-[43] ment Models. Journal of Derivatives, 1995, 3(2): 73-84.
68第 5 卷 第 1 期 澳 門 科 技 大 學 學 報 Vol.5 No.12011 年 6 月 30 日 Journal of Macau University of Science and Technology June 30 , 2011 收稿日期：2011-3-1；修訂日期：2011-5-15李威朗，男，澳門科技大學旅遊學院會展專業碩士畢業，澳門大學工商管理學士畢業。主要研究方向：社區參與及居民感知。本文在澳門科技大學國際旅遊學院唐娟助理教授悉心指導下完成。E-mail:Vincent_uv@yahoo.com.hk基於社區參與的澳門格蘭披治大賽車發展策略研究李威朗 （澳門科技大學國際旅遊學院，澳門）摘要：澳門格蘭披治大賽車是本澳旅遊發展業中的重要節事，已經成功舉辦了56屆。隨著賽車活動的定期舉辦，賽車發生地的居民負面影響日益凸顯。本文創新性地引入社區參與的理論視角，首次對澳門格蘭披治大賽車社區參與重大節事活動進行分析，在前人對澳門格蘭披治大賽車居民感知研究的基礎上，縱深了該領域研究空間，進一步拓展了若干個社區參與的研究主體。本研究通過文獻綜述對前人提出的社區居民負面影響感知進行再確認，然後用Delphi法彙集專家策略，並針對社區參與的最大群體做認同度問卷調查，問卷中設置開放性題目，使得居民可以發出自己的聲音。最後提出專家和居民共同支持的賽車負面影響解決策略，并提出激勵社區參與、提升參與層次的發展策略，以及構建社區參與保障機制，為澳門格蘭披治大賽車社區參與重大節事活動未來發展提出了建議，以供學界業界參考。關鍵詞：社區參與，澳門格蘭披治大賽車，發展策略A Study on Grand Prix Development Strategies Based on the Community Participation of the Macao ResidentsWai Long LEI( Faculty of International Tourism at the Macau University of Science and Technology, Macao )Abstract: The Macao Grand Prix is an important festival event in the Macao tourism industry, which has been held 56 times successfully. With the holding of regular racing, racing negative impacts on the Macao residents has become serious. This study innovative used community participation theory, for the first time analysised Macao Grand Prix community participation in activities of important events. Based on the former researchers of the Macao Grand Prix perception of the residents, deepening research space in this area, expanding a number of community participation subjects. This study re-confirmed the negative perception by the former researchers, through the literature review of previous negative effects of community, and then used Delphi method to collect strategies, and designed a survey questionnaire which have some open questions, so residents can make their voices heard. Finally, proposed the support solving the negative impact of Grand Prix strategies by both the residents and experts. In addition, proposed the strategies which could be encouraged community participation and enhanced the participation level of development. Furthermore, mechanisms involved in protection of the community, the Macao Grand Prix's future development strategies and recommendations for reference.Keywords: Community Participation; Macao Grand Prix; Development Strategies
71李威朗 基於社區參與的澳門格蘭披治大賽車發展策略研究形影響是難以衡量的，包括對社會生活和對團體福利的影響、事件所激發的自豪感，以及對一個地方或旅遊目的地形象所造成的長期影響，並認為舉辦重大事件活動的目的之一是塑造和延伸形象，特別是針對目的地區域或吸引主題來塑造形象，以此來扭轉原來的負面形象。Getz(1997)也認為，在無形影響中，重大事件活動對舉辦地形成的形象影響是重要的表現之一。周常春、戴光全（2005）通過對1999年昆明世博會的形象影響研究，將形象影響作為獨立的要素進行了研究，指出大型節事活動提高了舉辦地的知名度，有助於形成較高的形象認同度，並通過瞭解遊客的重遊意願和口碑效應來衡量長期的、潛在的形象影響。旅遊目的地居民感知研究回顧及研究現狀113 旅遊目的地居民感知研究回顧11311 隨著旅遊業的不斷發展和可持續發展理論提出，國外上世紀70年代出現了關於旅遊地居民態度和感知的研究，主要內容包括：居民對旅遊發展帶來的影響的感知、居民態度和感知的影響因素、居民感知和態度的相關解釋理論，以及以居民感知和態度的差異來將居民分類等方面（張欣然，2009）。我國的旅遊地居民感知和態度的研究則開始於上世紀90年代，研究的內容方法都逐漸增多。總的來看，國內外對於居民感知及態度的研究主要集中在以下四個方面：居民對影響感知的內容、影響居民感知的因素分析、基於居民感知差異的居民聚類分析、相關理論與研究方法。 澳門格蘭披治大賽車影響居民感知研究現狀11311在澳門眾多的節慶活動中，格蘭披治大賽車屬於享有最高國際聲譽的一項體育賽事。如今澳門格蘭披治大賽車已搖身一變，成為世界頂級車手爭相參與的一項車壇盛事。每年都會吸引到超過300名的賽車手及成千上萬的遊客。因此，格蘭披治大賽車是本澳旅遊業發展中的重要節事（李璽，2009）。為了格蘭披治大賽車的健康持續發展，澳門本地的相關機構及高校學者多年來進行了多項研究，主要有：2009年澳門科技大學周勇助理教授主持的關於大賽車居民感知方面的課題研究Resident's Perceptions towards the Impacts of the Macao Grand Prix; 2008年澳門旅遊學院進行的訪澳遊客及本澳居民對澳門格蘭披治大賽車的意見調查；澳門科技大學李璽助理教授發表的《澳門格蘭披治大賽車行銷策略研究》，為了該賽事能夠產生更大的帶動性和影響力從調查觀眾的消費行為出發來研究賽車的行銷策略。研究方法與研究設計1 研究方法111 本研究旨在探討如何降低格蘭披治大賽車對澳門居民產生的各種負面影響，應對居民感知和社區參與的變化規律、最大程度地降低賽車活動對居民生活的負面影響，建立社區居民的主人翁意識和尋找共同的利益和興趣，達成社區共識，共同為澳門格蘭披治大賽車未來發展獻計獻策。本研究採用文獻研究法、Delphi法、居民問卷調查法結合的研究途徑，來進行資料收集工作。簡單來說，通過文獻綜述和瞭解前人研究的基礎上，設計專家訪談問卷和居民調查問卷，保證問卷中所涉及的因子來源於可靠的文獻研究；Delphi法主要是用來收集專家、學者和業界權威意見建議，居民問卷調查則是用來瞭解居民的內在世界對專家訪談過程中收集到的策略持怎樣的態度。再輔以二手文獻資料，就可以提升研究結果的可信程度和解釋效度。文獻研究11111 由於格蘭披治大賽車是本澳旅遊業發展中的重要節事，且已經成功舉辦了56屆，故有不少的政府機關資料、國內外期刊論文、澳門學者專家研究、本地報章雜誌報導等資料來作為本研究之論述基礎。在前人考察社區居民感知和態度的變化規律的基礎上，以現有學者對居民感知的研究成果為起點，展開進一步的研究，填補前人研究力度相對較為薄弱的方面，以現有學者對居民感知的研究為基礎，拓展了若干個社區參與的研究
72主體，以體現對社區參與的重視。Delphi法11111 本研究以居民的觀點和視角切入，對大賽車活動行銷進行更寬廣與深入的研究。而筆者亦是澳門本地居民，多年來參與大賽車活動，對整個賽事和居民反映有較為直觀的瞭解，方便收集被研究者行為和態度的整個動力過程並以此設計具有針對性的訪談內容，利用Delphi法收集該領域專家學者針對降低居民負面影響、賽車未來發展方面提出的相關策略。居民問卷調查11113 居民問卷調查法則是在中和不同專家的立場之後，用以驗證Delphi法的結果並瞭解居民的內心世界對專家所提出之策略的態度。本研究為了想瞭解研究結果的效度，擬採用居民問卷調查的方式調查澳門居民，對研究結果和策略做補充與修正。本研究採非隨機抽樣的方法尋找調查對象，預計發放500份調查問卷。運用非隨機方便抽樣的方法，調查範圍覆蓋不同年齡、職業和教育背景，同時可以延展調查的地理範圍以及通過深入引導的提問得出進一步的資訊。研究設計111 首先，建立在對目前大賽車活動中居民感知研究的基礎上，本研究使用了Delphi法，分別在2010年7月到8月間，以及2010年9月到10月間，發送了兩輪專家訪談。採取計劃抽樣的方式，選取11位專家組成專家小組。這些專家至少符合下列條件之一：擔任高校旅遊專業的授課或研究工作達五 ●年以上；在民間社團擔任中層以上職務達五年以 ●上；任職於相關部門或機構，擔任中層以上職 ●務達五年以上；表1是本研究中Delphi法訪談對象的基本情況：表1 Delphi專家小組名單編號 現職 與本研究的相關性1 澳門科技大學國際旅遊學院副教授會展活動研究專家，曾主持澳門格蘭披治大賽車行銷策略課題研究。2 澳門科技大學國際旅遊學院助理教授 旅遊學院資深教授，會展發展策略研究專家。3 澳門科技大學國際旅遊學院助理教授會展活動研究專家，曾主持關於大賽車居民感知方面的課題研究Resident's Perceptions towards the Impacts of the Macao Grand Prix.4 澳門科技大學國際旅遊學院助理教授 會展活動研究專家，多年從事節事活動質量研究。5澳門品質管理協會會長；北京師範大學MBA客座教授。策略研究專家，為澳門眾多大型活動提供意見諮詢及可行性發展策略。6 澳門街坊會聯合總會副主任任職機構與賽車主辦單位合作舉辦賽車嘉年華活動；任職部門亦受理居民對於澳門大型節事活動的各種意見及投訴。7澳門街坊會聯合總會副理事長；澳門舊區重整諮詢委員會委員；旅遊發展輔助委員會委員；澳門特別行政區政府社區服務諮詢委員會中區召集人。任職機構與賽車主辦單位合作舉辦賽車嘉年華活動；任職部門亦受理居民對於澳門大型節事活動的各種意見及投訴；大賽車活動籌辦負責人之一。8 民間機構（家庭及青少年服務範疇）副主任 負責居民事務的處理與協調。9 民間機構（學校範疇）行政主任 負責居民事務的處理與協調。10 民間機構（青年範疇）資深社工 負責居民事務的處理與協調。11 復康機構主任 反映特殊群體的聲音。資料來源：本研究自行整理
73李威朗 基於社區參與的澳門格蘭披治大賽車發展策略研究在對第一輪專家意見匯總後，對各專家意見進行統計分析，綜合成新的訪談提綱，並把它分別再寄送給各位專家，由專家們對新訪談提綱做出第二輪判斷或打分。第一輪專家訪談採取半結構式的深度訪談來進行資料收集，提綱的設計主要為兩部份，以開放性題目為主，請專家為居民對大賽車影響較為反感的方面打分，並徵詢專家就該方面的對策和建議。其中，第一部份為受訪者的個人資訊，第二部份為主體資訊部份，對於文獻綜述中總結出來的典型性居民負面感知（交通擠塞 、環境污染、噪音擾民、公眾地方擠擁）再次進行徵詢核實，並進一步以開放式問題詢問改進的對策。第二輪訪談內容是建立在對第一輪統計匯總的基礎上，主要請專家對第一輪得到之策略和建議打分，採用利克特量表五分法，1分=非常不同意，5分=非常同意，以此類推，通過評分進行甄選，從中發現兩輪專家意見是否存在差異及產生差異的原因，最大限度地發揮群策群力的優勢，挖掘更多可行的策略。 然後，將專家提出的對策建議提供給居民，由居民以打分的形式對策略的支持程度進行評價。問卷主要分為四部份：第一部份為選擇性題目，分別考察居民的活動參與方式、參與程度、參與動機；第二部份為認同性評估題目，要求居民對每條策略的有效及支持程度進行打分：5分代表“非常同意”，1分代表“非常不同意”，以此類推；第三部份為開放性題目，如居民對澳門格蘭披治大賽車未來發展策略還有哪些想法和建議，可以選答該題；第四部份為受訪者基本資料，設計六個選項，包括：受訪者性別、年齡、受教育程度、職業、月收入、獲得賽事訊息的途徑。2010年10月19日～22日期間，筆者邀請30位澳門居民參加本研究的試調查，並根據文獻研究和試調查的結果，對問卷進行修改和完善，形成正式的居民調查問卷。筆者和同學及友人組成一行7人的調查團隊，分兩個階段展開調查工作，調查期間主要採用方便抽樣的方法，由調查員逐題訪問並根據受訪者的回答代為填寫，遇到受訪者不理解的問題調查員可以及時講解，這樣最大限度地保證了問卷的真實性和可信度。第一階段在2010年10月25～11月4日期間，調查組前往居民區進行街頭調查；第二階段在2010年11月5日～21日，調查組前往明愛園遊會和美食節嘉年華現場的澳門格蘭披治大賽車活動宣傳櫃檯和遊戲攤位，這些賽車攤位聚集了對賽車活動感興趣的大量澳門居民，為本調查的開展提供了方便。調查共發放問卷545份，得到有效問卷530份 ， 回 收 率 達 9 7 . 2 % ； S P S S 信 度 分 析 結 果 為Cronbach's Alpha = 0.8852，表明問卷題項的設計是可靠的。這樣的研究設計主要出於以下兩方面的考慮：一方面，出於對策略科學性的考慮，專家在相關研究或相關工作方面有著不可替代的優勢，他們對賽車負面影響及未來發展方面的認識具有長遠性和科學性的特點；另一方面，吸取專家的意見和策略，不代表輕視居民的智慧與聲音，相反，在專家訪談之後，將專家意見拿給居民進行打分，正是體現對旅遊社區的最主要利益群體——居民的重視。數據分析3 資料收集的分析是用統計套裝軟體SPSS17.0版本。主要運用了平均數來做對比，分析了策略方面專家打分和居民打分之間的差異。對於專家訪談所總結出的負面感知解決對策以及提出的澳門格蘭披治大賽車未來發展策略，居民的總體認同度平均值為3.8813分，趨向於“認同”，可以認為居民對專家策略的總體認同性傾向於正面評價。專家策略與居民認同性分析統計結果如表2。在和專家打分進行對比後，獲得每條策略的專家打分與居民打分的平均差，使用二者打分平均值差異詮釋每一條策略的個體認同度。
81李威朗 基於社區參與的澳門格蘭披治大賽車發展策略研究建立利益保障機制11111 （1）完善補償制度制定相應的規範，完善補償制度，確保澳門社區居民的利益有據可依。成立專門機構保證社區接待居民的意見和投訴，最終保證社區參與權力的實現。在特區政府的大力支持下，設立補償基金，專門用於彌補節事活動舉辦給社區居民帶來的負面影響和損失。例如，實行受影響的區域內公交線路免費，並且增加區域內公交線路的運作時間，緩解交通方面的負面影響；進行噪聲擾民程度的實踐訪談和調查，對於受噪聲影響較為嚴重的片區的居民、社團組織，採取一定的補償措施：經濟補償、免費發放若干賽車入場門票等；可以為受噪聲影響較為嚴重的居民家庭安裝隔音玻璃和派耳塞給受影響區域的居民等。（2）利益分享制度利益分享制度目的在於使整個社區和社區內的居民都能在活動中有所獲益。例如成立賽車知識推進委員會，在居民、小朋友中進行趣味知識宣傳、培訓、有獎問答，為有相關領域興趣的居民提供免費知識培訓、乃至親身試駕等體驗性活動，讓居民能從活動中有所斬獲；或者將格蘭披治大賽車活動與公益活動相結合，為整個澳門社會做出自己應有的貢獻等。建立監控機制11113 可以成立由澳門居民組成的賽車活動監督機構，保證他們參與格蘭披治大賽車的全過程，確保居民賽車活動帶來的各種利益，也可以監督旅遊管理部門和主辦機構相關經營行為，確保社區居民的整體利益不被損害。建立評估機制11114 由澳門特區政府旅遊局組織旅遊專家制定評估標準，其過程要充分聽取社區居民的意見和建議。專家根據澳門格蘭披治大賽車的實際情況，確立一套評賽車活動影響評估標準。然後通過問卷調查、座談會等方式廣泛吸收社區居民反映的經濟、生活和環境狀況，並將其意見和建議進行分析，在結合對賽車活動的實際考察，作出評估。使賽車活動的影響及影響範圍評估可以更加科學化、實際化。建立反饋機制11111 回饋機制應貫穿賽車活動的整個過程，鼓勵澳門居民積極發表自己對格蘭披治大賽車發展的想法和看法，尤其是涉及居民自身利益的方面，自由表達意願。例如可以定期舉行居民座談會，設立專門機構、專用信箱、郵箱、熱線電話，或採用定期走訪、調查問卷等方式，保證澳門居民可以及時有效回饋意見和問題。對回饋的意見及建議及時整理，並作綜合分析，結合實際情況，邀請專家做科學評價，接納其中合理正確的意見及建議，並把不合理缺乏科學性的問題及時向居民解釋清楚。參 考 文 獻Burgan B . & Mules T. (1992). Economic impact of Sporting Events. [1] Annals of Tourism Research, 19(4), 700-710.Cevat Tosun. (2000). Limits to Community Participation in the [2] Tourism Development Process in Developing Contries. Tourism Management, 21(6), 613-633.Getz, D. (1997). Event Management & Event Tourism. New York: [3] Cognizant Communication Corporation.Murphy P. E. (1985). Tourism: a community approach. Methuen: New [4] York and London, 176.Petty J. (1995). The many Interpretations of Community Participation. [5] In Focus, 16, 5-10.蓋宏君，周靜言（2007）。事件旅遊對城市旅遊發展的影響，經[6] 濟研究導刊，6，189-191.蔣德恩（2006）。節事旅遊在加拿大旅遊業發展中的作用，北京[7] 第二外國語學院學報，3，25-26.李璽（2009）。澳門格蘭披治大賽車行銷策略研究，澳門研究，[8] 52，137-140.羅秋菊（2003）。世界大型事件活動對旅遊業的影響極其對中[9] 國的啟示——以歷屆奧運會和韓國世界盃為例，商業研究，11，150-152.約翰·艾倫等（2002）。大型活動專案管理。北京：機械工業出[10] 版社（26-27頁）.曾艷（2007）。國內外社區參與旅遊發展模式比較研究（碩士論[11] 文）。取自中國知網資料庫.周常春，戴光全（2005）。大型活動的形象影響研究——以昆明[12] 世博會為例，人文地理，2，38-43.張麗，郭英之（2005）。重大事件的旅遊效應分析及行銷策略，[13] 商業研究，18，70-171.張欣然（2009）。國內外旅遊影響和居民感知及態度研究綜述，[14] 旅遊研究與發展，11，80-81.鄭向敏，劉靜（2002）。論旅遊業發展中社區參與的三個層次，[15] 華僑大學學報，4.
82第 5 卷 第 1 期 澳 門 科 技 大 學 學 報 Vol.5 No.12011 年 6 月 30 日 Journal of Macau University of Science and Technology June 30 , 2011 收稿日期：2011-2-2；修訂日期：2011-2-18梁文慧博士是澳門科技大學持續教育學院總監兼國際旅遊學院院長與旅遊管理教授。 論文為澳門基金會資助研究專案 “兩岸四地成人遙距自我調控學習素養之比較研究＂的階段性成果。專案成員包括澳門科技大學梁文慧教授、臺灣高雄師範大學王政彥教授、李嵩義助理教授和龔雙慶研究助理等。香港遙距學習者自我調控學習素養培養 的困難和對策梁文慧 （澳門科技大學持續教育學院/國際旅遊學院）摘要：論文旨在調查中國香港地區在高等院校進行遙距學習的學生在遙距自我調控學習素養的現況，希望藉由訪談的方式，將搜集的資料、統計的分析、訪談的結果，以瞭解香港地區的遙距自我調控學習素養，並為日後針對結果之異同加以比較， 提出具體的建議，用作提升香港的成人遙距學生的學習效果，並為增進成人自我調控學習的成效提供優化策略的參考。關鍵詞：自我調控學習；遙距學習；學習素養；香港Cultivating Online Self-regulated Learning for Adult Learners in Hong Kong: Its Barriers and CountermeasuresAliana Man Wai LEONG( Faculty of International Tourism/the School of Continuing Studies Macau University of Science and Technology, Macao, China )Abstract: This study aims at exploring the current condition of adult online self-regulated learning and its effect on adult literacy in Hong Kong. It is expected that through interview, data collection and statistical analysis to understand adult online self-regulated learning and its effect on adult literacy in Hong Kong and compare similarities and differences of the results to build concrete proposals to make adequate preparations to enhancing the effectiveness of adult self-regulated learning in Hong Kong.Keywords: Self-regulated learning; Online learning; Learning literacy; Hong Kong引言0 本研究為澳門基金會資助專案 “兩岸四地成人網路自我調控學習素養之比較研究”的階段性研究成果。旨在探討自我調控學習的理論並加以分析，然後統整相關的實證研究敍述，從而為研究項目勾劃出堅實的理論基礎。然而，整個研究時間將尋求調查中國內地、香港、澳門及臺灣四個地區成人遙距自我調控學習素養的現況，希望藉由問卷調查及分區訪談的搜集資料及統計分析、訪談結果的呈現，瞭解中港澳臺四地區透過遙距學習的成人學習者自我調控學習素養的內涵，並根據調查結果加以比較異同並提出具體的建議，促進中國大陸、香港、澳門及臺灣地區的成人學習者在遙距學習效果的提升與增進成人網路自我調控學習成效。本研究意義與價值非常深遠，他山之石可以攻玉，跨國(地區)比較研究是與全球接軌的途徑，
89梁文慧 香港遙距學習者自我調控學習素養培養的困難和對策應定期開辦理相關研習活動，並制定遙距 ●教育之教師有關進修制度及辦法。由遙距學習所得之文憑及學歷，其認證應 ●採較寬的辦法，以強化在職進修的成人學習者對於遙距教育的興趣。參 考 文 獻Boekaerts, M. (1999). Self-regulation learning: Where we are today? [1] International Journal of Educational Research, 31, 445-447.Cheung, C., & Kwok, S. (1998). Activities and academic achievement [2] arming college students. Journal of Genetic Psychology, 159,147-162.Corno, L. (1989). Self-regulated learning: A volitional analysis. In B. J. [3] Zimmerman & D. H. Schunk (Eds.), Self-regulated learning and aca-demic achievement: Theory, research, and practice (pp.83-110). NY: Springer-Verlag.de Jong, F. P. C. W. (1990). Self-regulation trained in mathematics. In [4] J. M. Pieters, K. Brever, & P. R. J. Simons (Eds.), Learning environ-ments: Contributions from Dutch and German research (pp.299-303). Berlin: Spring Verlag.Everson. H. T. et al. (1997). Do metacognitive skill and learning strat-[5] egies transfer across domains? Paper presented at Annual meeting of the American Educational Research Association, March, 24-28. Chi-cago.Harter, S. (1982). A developmental perspective on some parameters of [6] self-regulation in children. In Karoly, P., & Kanfer, F. H. (eds.), Self- management and Behavior Change: From Theory to Practice (pp.165-204). New York: Pergamon. Lan, William Y. (1996). Self-regulated learning strategies of inter-[7] national students: a study of high and low achievers. The Journal of Experimental Education, 64, 101-115.Lidner, R.W., & Harris, B. R. (1993). Teaching self-regulated learning [8] strategies. In M.R. Simonson and K. Abu-Omar (Eds.). Proceedings of Selected Research and Development. Presentations at the 1993 Na-tional Convention of the Association for Educational Communications and Technology, New Orleans, LA. Ames, IA: AECT, 641-654.Lindner, R. W. & Harris, B. R. (1992b). The development and evalu-[9] ation of a self-regulated learning inventory and its implications for instructor-independent instruction. Proceedings of selected research and development presentations at the Convention of the Association for Educational Communications and Technology, Iowa, January 15-18.Lindner, R. W. et al. (1996). The design and development of the “self- [10] regulated learning inventory: a status report. Paper presented at the Meeting of the American Educational Research, New York (ERIC Document NO. ED 401 321).Mace, F. C., Belfiore, P. J., & Shea, M. C. (1989). Operant theory [11] and research on self-regulation. In P. J. Zimmerman, & D. H. Schunk (Eds.), Self-regulated learning and academic achievement：Theory, research, and practice (pp27-50). NY: Springer-Verlag.McCombs, B. L. (1986). The role of the self-system in self-regulated [12] learning. Contemporary Educational Psychology, 11, 314-332. Morris, D. R., Gredle, M. E., ＆ Schwartz, L. S. (1998).The role of [13] self-regulated learning in industrial training environment. Paper pre-sented at the Annual Meeting of the American Educational Research Association. April, 13.17. San Diego, Ca.Pintrinch, P. R. & DeGroot, E. V. (1990). Motivational and self-[14] regulated learning components of classroom academic performance. Journal of Education, 82(1), 33-40.Pintrich, P. R. (1986b, April). Motivation and learning strategies inter-[15] actions with achievement. Paper presented at the American Education Research Association Convention, San Francisco, California.Pintrich, R. R., Smith, D. A. F., Garcia, T., & McKeachie, W. J. (1993). [16] Reliability and predictive validity of the motivated strategies for learning questionnaire (MSLQ). Educational and Psychological Mea-surement, 53, 801-813.Pintrinch, P. R. & DeGroot, E. V. (1990). Motivational and self-[17] regulated learning components of classroom academic performance. Journal of Education, 82(1), 33-40.Pokay, P., & Blumenfeld, P. C. (1990). Predicting achievement early [18] and late in the semester: The role of motivation and use of learning strategies. Special Section: Motivation and efficacy in education: Re-search and new directions. Journal of Educational Psychology, 82(1), 41-50.Pressley, M., Ross, K. A., Lewin, J. R., Ghatala, E. S., & Ahmad, M. [19] (1987). Test monitoring in young grade school children. Journal of Experimental Child Psychology, 43, 96-111.Schuitz, P. A. L. (1997). Educational goals, strategies use and the [20] academic performance of high school students. High School Journal, 80(3), 193-202.Schunk, D. H. (1996). Learning theories: educational perspectives [21] (2nd Ed.). Englewood Cliff, New Jersey: Prentice-Hall, Inc.Stoynoff, S. (1996). Self-regulated learning strategies of international [22] students: A study of high and low achievers. College Student Journal, 30(3), 329-336.Trawick, L. (1988). Relationship among cognitive-motivational pro-[23] cesses and academic performance in community college student with a history of academic failure. Paper presented at annual meeting of the American Education Research Association, New Orleans, LA. Zimmerman, B. J., & Martinez-Pons (1986a). Development of a [24] structured interview for assessing student use of self-regulated learn-ing strategies. American Educational research Journal, 23(4), 616-628.Zimmerman, B. J., & Martinez-Pons (1988). Construct validation of [25] a strategy model of student self-regulated learning. Journal of Educa-tional Psychology, 80(3), 284-290. Zimmerman, B. J., & Martinez-Pons, M. (1990). Student difference [26] in self-regulated learning: Relating Grade, sex, and giftedness to self-efficacy and strategy use. Journal of Educational Psychology, 82, 51-59.梁文慧、王政彥（2007）.兩岸四地大學發展持續教育合作研[27] 究。中國：清華大學出版社，ISBN 978-7-302-15324-5.王政彥（2002）.自我調控學習在成人高等教育機構教學上的應[28] 用，成人教育，69，20-28.附 錄 一 訪 談 大 綱(一) 從研究早期發現得知，香港地區成人在遙距的學習過程的改進上不及中國大陸及臺灣地區，您認為此差異的可能原因為何？您認為香港地區成人遙距學習者學習過程改進的具體策略為何？
91第 4 卷 第 2 期 澳 門 科 技 大 學 學 報 Vol.4 No.22010 年 12 月 30 日 Journal of Macau University of Science and Technology Dec 30 , 2010 收稿日期：2010-10-17；修訂日期：2011-1-27本文系澳門基金會資助專案《澳門宗教旅遊資源挖掘——澳門賭神形象塑造與開發》﹝專案代碼：0089﹞的研究成果之一。課題負責人：澳門科技大學葉桂平博士。作者簡介：王子超，澳門科技大學國際旅遊學院助理教授，武漢大學中國傳統文化研究中心歷史學博士，主要從事旅遊文化方面的研究。王克陵，武漢大學歷史地理研究所教授，主要從事歷史景觀地理及其文化研究。中國傳統賭神崇拜試探王子超1，王克陵2（1. 澳門科技大學國際旅遊學院，澳門；2. 武漢大學歷史地理研究所，武漢）摘要：本文分析了中國古代人間真實的博戲活動和頗為流行的博戲神話，研究中國賭神崇拜文化產生與變化的歷史。探討“博弈＂當中，“不確定性＂基礎上產生的賭神崇拜文化特徵，並分析其人間原型之不同背景與心態。根據博弈的文化心理流變和西嶽金天王崇拜典型，試解人們為甚麼在實利的博弈中，拜祭那執掌“不確定性＂結果的神祗，從而探索中國傳統文化中賭神崇拜的深刻淵源。關鍵詞：賭神崇拜；博戲；神話 Original Culture Research of God of Gambling Belief of ChinaZichao WANG1, Keling WANG2(1. Faculty of International Tourism, Macau University of Science and Technology, Macao, China)(2. Wuhan University, Wuhan, China)Abstract: The God of gambling belief is a common folk phenomenon. Before entering the casino, many people would like to burn incense in Macau’s temples,praying for super power to win in the coming gambling. At the angel of God of Gambling belief in Macau, this paper studies the real and ancient gambling activities and some famous fairy tales of ancient China,trying to find the evolutive history of the belief in China.Firstly, this paper studies the cultural features of uncertainty and God power. Secondly, this paper analyzes different background and considerations of people with worship of God of Gambling. Thirdly, according to the evolution of cultural psychology and the typical worship of Golden King of China’s Huashan Mountain, the paper tries to find the reasons of praying for God who powers the uncertain results in gambling activities for actual benefit. In summary, we try to explore the origin of God of gambling belief in the traditional culture of China.Keywords: God of Gambling Belief; Gambling; Fairy Tales前言0 中國的博戲文化與神仙文化，在複雜的歷史、經濟文化發展過程中形成了一種特殊現象。不少博戲活動參與者有著心中的賭神並加以崇拜，祈求在博弈中獲得好運，可稱為賭神崇拜。我們試對這一文化現象的源流和動因作深入分析。中國古代文獻中很早就記載了“博弈”。《論語·陽貨》載：“飽食終日，無所用心，難矣哉！不有博弈者乎？為之，猶賢乎已。[1]”玩一
98第 5 卷 第 1 期 澳 門 科 技 大 學 學 報 Vol.5 No.12011 年 6 月 30 日 Journal of Macau University of Science and Technology June 30 , 2011 收稿日期：2010-11-8；修訂日期：2011-1-6張雁，女，南京大學文學博士。澳門科技大學通識教育部助理教授。主要研究方向為中國古代文學及文獻學。王端淑與《名媛詩緯初編》張 雁（澳門科技大學通識教育部，澳門）摘要：晚明時期所出現的大量女性詩歌總集、別集與選集，昭示了當時女性文學創作的繁榮與興盛。明末女詩人王端淑編纂《名媛詩緯》收錄了近千家明代女性詩人作品，洵為明代女性作品之大觀。王端淑在編纂收錄女性作品的過程中，闡發了她的文學發展觀和審美觀，並從性別角度對女性文學創作的特點進行了總結與概括，勾畫了明代女性文學創作的歷程和圖景。關鍵詞：女性文學創作；王端淑；《名媛詩緯初編》Duanshu Wang and The Preliminary Selection of Ladies’ PoemsYan ZHANG( Department of General Education, Macau University of Science and Technology, Macao, China )Abstract: A lot of collections of ladies’ poems were printed in the later period of Ming Dynasty. The Preliminary Selection of Ladies’ Poems edited by Duanshu Wang was famous for its considerable amount. In this selection, Wang illustrated her aesthetic views on literature, described the characteristic and the course of feminist literature of Ming Dynasty.Keywords: Feminist literature; Duanshu Wang; The Preliminary Selection of Ladies’ Poems前言0 明代以來，伴隨著女學教育的推廣，印刷的普及和心學的勃興，女性文學活動日趨興盛。大量女性文學作品的出版，一方面證實了女性文學創作力的提昇和創作的活躍；另一方面也突顯了當時社會對於女性文學作品的欣賞與喜愛。隆慶四年（1570）俞憲編選《盛明百家詩》時，云《淑秀總集》“十七人，共詩、賦、詞七十二首”之編纂，只是為了“以備典故，庶後之覽者有考云”[1]，“明非所重，亦只以異耳”[2]。至萬歷、崇禎年間，女性的才能不斷得到男性士人的肯定和褒獎，“非以天地之靈秀之氣，不鐘於男子；若將宇宙文字之場，應屬乎婦人”[3]。書商們對於市場風向的準確把握直接催生了晚明一大批女性作品選集、總集，如周履靖《古今宮閨詩》、《香奩詩》、鄭文昂《名媛匯詩》、趙世傑《古今女史》、胡文煥《新刻彤管摘奇》，蘧覺生《女騷》等，但需要注意的是，書商們編纂的急就章往往在內容上互相重複勦襲，體例復加舛謬，且好請著名文士作序為之闡揚，誇飾作品選輯之豐富以及女性作品之盡善盡美。如趙世傑《古今女史序》：海內靈秀，或不鍾男子而鍾女人。其稱靈秀者何？蓋美其詩文及其人也……雖曰宇宙寥廓，世代緜邈，而女郎之輕俊，饒有聲澤者，項背相望於世。大都外觸於境，而內發於情，指冷齒芬，香喉
104第 5 卷 第 1 期 澳 門 科 技 大 學 學 報 Vol.5 No.12011 年 6 月 30 日 Journal of Macau University of Science and Technology June 30 , 2011 收稿日期：2010-11-18；修訂日期：2010-12-29通訊作者：彭勃，男，澳門科技大學法學院2009級博士研究生，深圳大學法學院教授，日本一橋大學法學博士。主要從事刑事訴訟法學、比較法學的研究。Email：pengbo@szu.edu.cn. Tel:+86-75526534437.刑事和解之定位重塑與制度出路彭 勃（深圳大學法學院，深圳）摘要：在我國刑事訴訟領域，正在推行“刑事和解＂程式，與之相關的理論爭議也隨之產生。由於對該制度的概念、淵源及制度價值性等基本問題尚無法形成共識，實際操作中產生了制度定位不清、操作程式不規範等弊端。本文針對“刑事和解＂的概念、制度淵源、理念基礎以及適用範圍之主流觀點進行了梳理，通過與恢復性司法、辯訴交易等相關制度進行比較，指出該制度的新定位，即不僅應實現與現行法制的對接，還應構建一種全新的官民合作型糾紛解決機制。同時，對於“刑事和解＂制度出路問題，本文認為應當以恢復性司法的最大化模式為借鑒，將“刑事和解＂改造成一種以實現相對正義為目標的開放型、包容型司法制度。關鍵詞：刑事和解；定位重塑；制度出路；替代性解決機制；恢復性司法；辯訴交易Reposition and Construction of Victim-Offender Mediation: Protection of Victims Rights under Restorative Justice PatternBo PENG( Shenzhen University Law School, Shenzhen, China )Abstract: The article fully discusses the core objective of the restorative justice pattern which focuses on the negotiation and restoration and deals with cases in conformity to the victims' wills in some degree. It's advantages are to enforce the protection of the victims and their interests, restore the relations between the parties concerned, and keep the social harmony and peace, despite some drawbacks. Therefore, the completely new concept and model are conducive to the protection of the victims' rights.Keywords: Victim-offender; Nediation; Restorative; Justice; Victim; Protection of rights引言0 改革開放30年來， 我國的政治、經濟和社會形勢發生了深刻的變化。市場經濟條件下的各種社會矛盾與問題日益突出，犯罪形勢嚴峻，發案數量持續上升，案件堆積如山。現行的刑事糾紛處理機制雖然積極應對，總體上發揮了積極的作用，但因體制單一、機制不順暢，導致司法資源浪費，訴訟效率低下。為貫徹落實科學發展觀、構建以人為本的法治環境，中共中央十六屆六中全會在《中共中央關於構建社會主義和諧社會若干重大問題的決定》中要求在刑事司法領域推行“寬嚴相濟的刑事司法政策”，以保障社會“和諧、有序發展”。以此為背景，近年來我國訴訟實踐部門開展了一些替代性的刑事糾紛處理方式，以期糾正傳統刑事司法體制下偏重犯罪控制、有罪必罰的流
109彭勃 刑事和解之定位重塑與制度出路委員會第11屆會議通過《關於在刑事事項中採用恢復性司法方案的基本原則》的決議，鼓勵各成員國在制定、開展恢復性司法時參照聯合國的標準。該決議採納了美國學者馬歇爾（T•Marshall）的觀點，認為“恢復性司法是在公正第三方幫助下，被害人、罪犯和（或）受犯罪影響的其他個人或社區成員共同參與解決犯罪問題的法律程式”[17]。然而，以瓦格雷夫（L•Walgrave）為首的另一派學者則賦予恢復性司法更為寬泛的概念外延，即“針對犯罪危害而開展的一切修復性法律行為都屬於恢復性司法。[18] ”為了對這兩種定義加以區別，學界一般稱前者為“純粹模式”（Purist Model），稱後者為“最大模式”（Maximalist Model）[19]。“純粹模式”主張將受犯罪影響的當事人組織起來，通過參與、對話和協商，促使罪犯認識其犯罪行為給他人和社會所造成的危害，從而實現犯罪人、被害人和其他當事人的和解。在這一模式中，犯罪被認為是對他人和社會和諧關係的一種破壞，因此各方當事人的直接參與合作被認為是解決糾紛不可或缺的條件。“純粹模式”反對運用暴力和強制措施，強調糾紛解決的自願性，推崇以溫和的“會商”與“面談”來取代刑罰，體現了一種樸素的理想主義司法觀。在“純粹模式”的支持者看來，成功的恢復性司法運作是各方當事人共同努力的結果，必須保證各方當事人都能參加與合作，因此推出了“被害人與加害人調解”（ Vi c t i m -Offender Mediation[VOM]）、“家庭小組會議”(Family Group Conference[FGC])、“社區會商”（Conferencing）、“圓桌討論”（Circles）等操作形式[20]。修補因犯罪而受損的加害人、被害人和社區的關係、促使加害人反省其行為的社會責任以及提供社會援助是“純粹模式”的三大特點。可見，“純粹模式”意圖在加害人、被害人和社區之間構建直接會商與利益平衡的機制，以修補被破壞的社會關係、實現被害人救濟及修復三方關係，並最終使犯罪人走上改過自新與複歸社會之路。不過，從法律效果和社會評價的角度來看，“純粹模式”的缺陷也是明顯的。（1）該模式要求受犯罪影響的當事人“齊聚一堂”進行直接對話，強調參與者的自願性，反對在程式中使用強制手段促成協商。然而，在很多時候要求當事人會商並非易事。由於犯罪帶來的痛苦經歷，雙方當事人往往對直接見面感到“不安”和“困惑”，雖然他們主觀上並不反對和解，也不拒絕參與恢復性程式[21]。（2）作為刑罰權國家壟斷的一種替代，“純粹模式”主張通過社區的主導作用來推動刑事糾紛的解決。因此要求社區（代表）必須參與全部會談過程，以修補因犯罪而受損的社會整體利益。儘管沒有人否認社區在解決糾紛、監督協議執行方面的積極作用，但也應充分注意到社區協調的局限性。在社會權力取代國家權力於糾紛解決機制中發揮作用的同時，這種權力有時也會超越合理的範疇，形成一種“集體暴政”[22]。因此，對於“純粹模式”下的社區主導原則理應保持必要的警惕，並設置相應的制約性程式。與之相對，“最大模式”並不要求必須做到“相關當事人共同參與”。在該模式之下，為了實現對被害人救濟與修復社會創傷，可以在恢復性司法中適用強制手段，以督促加害人參與和解活動。最大模式既關注糾紛解決方案的運行過程，也重視制度的實現效果。從這種意義上講，該模式更加務實和開放，不僅包含了純粹模式的基本內容，而且提供了更為豐富的操作方式。在“最大模式”下對於主動認罪、悔罪和不具有重新危害社會的嚴重犯罪加害人也可以成為恢復性司法的適用對象。與之對應，對於無心悔改或惡意利用恢復性司法程式拖延訴訟進程的加害人，國家權力機關有權對其賦課修復性制裁。目前，世界範圍內的恢復性司法實踐已不限於該制度的早期雛形——“家庭小組會議”（FGC）方式，更逐漸發展出多種制度類型。因而，純粹模式”與“最大模式”並非兩種對立的
113彭勃 刑事和解之定位重塑與制度出路人雙方自行協商、交涉的結果，國家的訴訟權力機關基本上不參與和解討論，只是根據雙方要求對協定進行審查和確認。不過，當雙方當事人達成了和解協定後，訴訟權力機關有權對和解協議的合法性或合政策性進行審查，並決定是否終止針對加害人的刑事追訴或者減輕其刑事責任。實踐中，“自行和解模式”多見於加害人與被害人積怨不深、加害人主動賠償的輕傷害案件或者熟人之間的輕微案件[34]。與之相對，“司法調解模式”是公、檢、法機關主動與加害方、被害方以及其他相關人員和單位進行溝通、交流、教育、勸解的解紛機制。在調解中，主持機關一方面要促使加害人真誠悔罪和道歉，並向受害人提供合理的賠償；另一方面也要說服受害人消解怨恨，理性求償，以利於雙方和解。在一些個案中，主持機關甚至要協助商定和解方案，才能促使雙方當事人會商。當然，儘管主持機關在調解中表現得比較主動，但是和解仍然是以加害人、被害人之間的合意或者默認為基礎的。此種調解模式主要用於加害方與被害方存在尖銳矛盾，加害人不願意主動和解，同時被害方又有著強烈復仇心態的案件[35]。“人民調解委員會調解模式”屬於一種委託確認模式。調解的參與者由國家訴訟權力機關、仲介機構(通常是調解委員會)、加害人和被害人四方組成；調解程式首先由訴訟權力機關委託相關仲介機構進行調解，雙方當事人達成和解後交由訴訟權力機關進行合法性或合政策性的審查及確認。在此種和解模式中，訴訟權力機關既具有主動性又不過度介入，和解結果同樣都取決於加害人與被害人的合意或同意。實踐中，“人民調解委員會調節模式”可適用於各種案件類型，對於前述兩種模式無法處理的案件也可做調解[36]。不過，無論基於何種劃分標準，人們理應注意到：（1）實踐中的“刑事和解”，大多同時具備幾種模式的特徵或者幾種模式並行適用，與理論模式完全吻合的“刑事和解”並不多見；（2）上述幾種和解模式並非新近才有的，在傳統的刑事司法活動中業已存在，只是並非被普遍接納，適用範圍也較為狹窄；（3）“刑事和解”並非一成不變，而是在不斷發展與轉型之中，因此應當以一種開放與耐心的姿態對中國“刑事和解”實踐狀況加以考察。 刑事和解的定位重塑313 根據以上分析，可以歸納出這種制度的幾個基本特徵：1）無論在何種“刑事和解”的制度模式中，國家訴訟權力機關都不再對犯罪的實體處理權實行全面壟斷，而是將部分實體處理權轉交給加害人、被害人或者第三方組織（如社區、人民調解委員會等）。2）國家訴訟權力機關雖然讓渡出部分實體處分權，但仍牢牢地掌握著對案件結論的終局性審查、確認及否決權；與之相對，案件的被害人、加害人通過參與“刑事和解”，獲得了一定程度的結果決定權，但這種決定權並非單方獨享，而是體現為雙方合意或默認和解。3）在“刑事和解”諸種模式下，對話、溝通、協商、妥協是和解過程中的重要手段，並形成了參與程式的各方博弈的格局。如果將具有上述特徵的“刑事和解”與傳統刑事司法制度相比較，不難發現兩者的本質差異。傳統的刑事司法活動具有兩個突出的特點：一個是刑事法制的國家中心主義。即認為犯罪侵犯的是國家利益，國家作為“和平的守護者”（keeper of the peace），有權“運用法律，確立罪責，實施懲罰。”國家的這種角色定位導致犯罪被害者被剝奪了作為訴訟當事人的法律地位，失去了參與刑事司法的機會。“刑罰權國家壟斷”的做法滿足了國家展示權力和權威的需要，被害人、社區等其他相關主體的利益則僅在抽象意義上(即由國家代表他們)予以考慮，而犯罪行為所造成的損害很少獲得補償。另一個是刑事法治的形式性，即以類型化和要件化的方式來描述犯罪，並以一種程式化的方式來進行犯罪處理。在這樣一種司法中，刑事案件被類型化為犯罪構成要件
119彭勃 刑事和解之定位重塑與制度出路保持“刑事和解”在刑事法治中的特殊定位，並在此基礎上真正發揮其制度潛能，或許該制度在中國得以確立與的前提，也是訴訟實踐機關和學術界都需要認真思考的問題。參 考 文 獻彭于 .最高檢正對各地刑事和解實踐開展調研[N].法制日報，[1] 2007-9-21(4).近年來媒體上也出現了質疑刑事和解的聲音，例如初殿清.檢視[2] 刑事和解制度的平等性質疑[N].檢察日報,2007-6-11(4).；李承錦.河南首例重罪案刑事和解後被輕判引質疑[N].工人日報,2009-10-26(5).陳瑞華：刑事訴訟的私力合作模式——刑事和解在中國的興[3] 起.中國法學，2006(5)20.中山大學法律評論[M],2010(8):307.[4] 參見中國人民大學張志銘教授在“寬嚴相濟刑事政策與刑事和解[5] 研討會”上的觀點，刑事和解的理論基礎：法律家與法學家對話錄，國家檢察官學院學報[J],2007(4）.陳瑞華.問題與主義之間:刑事訴訟基本問題研究，北京：中國人[6] 民大學出版社，2008（2).中國政法大學樊崇義教授認為,將刑事和解的起源理解為加拿大[7] 安大略州1974年的判例是一種誤解.中國博大精深的和合思想就蘊涵了和諧司法的理念,這種理念較之於恢復性司法理念,在內涵上更加全面和科學.和諧社會理念的提出以及和諧司法理念的建立在我國具有本土化的特點,有其深層次的社會和文化背景,具有強大的生命力.在刑事和解制度的構建上我們應該堅持中國特色.西北政法大學賈宇教授根據1941年7月21日延安地方法院判決書第10號的一個案例，也認為刑事和解起源於我國。黃京平、甄貞.和諧社會語境下的刑事和解,北京：清華大學出版社，2006.梁治平 .清代習慣法：社會與國家 ,北京：中國政法大學出版[8] 社,1996.20世紀70年代基陳納市開始進行“被害人與加害人”和解嘗試。[9] 當時，該市的一名年輕緩刑官員說服法官讓兩名被判處破壞藝術作品犯罪的年輕人同所有的被害人見面。其後，法官責令兩年輕人向被害人賠償所有損失作為其判處緩刑的條件。數月後，兩名加害人再次會見所有被害人並支付相應的賠償以履行法院判決。該市的這種嘗試逐漸演變為一個由教會捐贈、政府補助和社會各界支持的“被害人與加害人”和解基金會。隨後，加拿大其他地區也積極參與這項活動。1978年，美國印第安那州埃爾德哈特市首次將“被害人與加害人和解”方案引入美國。自此，該和解方案迅速傳遍了整個美國和歐洲。目前，世界上已出現了1200多個“被害人與加害人和解”項目，其中美國和歐洲占75%以上.林澤波.刑事和解制度之“源流考”與“法理辨”—刑法史學、[10] 刑法哲學與犯罪學視域下的探究.中國監獄學刊,2007(6):45.熊秋紅.刑事辯護論.北京：法律出版社，1998.[11] “嚴打”是依法從重從快，“嚴厲打擊刑事犯罪分子活動”的簡[12] 稱，是解決一定時期中突出的社會治安問題，而依法進行的，打擊嚴重刑事犯罪的活動。中國於1983年首次提出“嚴打”這個概念，並進行第一次“嚴打”。另外在1996年第二次“嚴打”和2000年至2001年進行還增加了網上追捕逃犯的“新世紀嚴打”.王永興.刑事和解的適用範圍研究—結合地方關於刑事和解規定[13] 的分析.福建員警學院學報,2010(2):95-102.最高人民檢察院副檢察長朱孝清在2008年12月7日舉行的“刑事[14] 和解與程式分流”研討會上公開表示，我國目前的刑事和解案件應僅限於加害人與被害人之間就精神撫慰、民事賠償達成的和解，且必須以侵害的是特定被害人利益、加害人一方真誠悔罪和雙方當事人自願和解為前提。轉引自李松，黃潔.最高檢 刑事和解不是“花錢買刑，共產黨員[J]，2009(1):36.截止到2005年，恢復性司法已在西歐、北美等80多個國家[15] 得 到 不 同 程 度 地 應 用 ， 世 界 範 圍 內 的 恢 復 性 司 法 項 目 多 達1000多個。恢復性司法模式在國際上得到了普遍的認可。Van Ness,An Overview of Restorative Justice around The World, Workshop2:Enhancing Criminal Justice Reform, including Restorative Justice, Paper at Eleventh United Nations Congress on Crime Prevention and Criminal Justice,pp1-27.2005. Declaration of Leuven on the Advisability of Promoting the [16] Restorative Approach to Juvenile Crime, in: Walgrave (ed.), Restorative Justice for Juveniles: Prtentialities, risks and problems for research, 1998, pp.389-401. 聯合國.關於在刑事事項中採用恢復性司法方案的基本原則，§2.[17] Bazemore/Walgrave, Restorative Juvenile Justice: in Research of [18] Fundamentals and an Outline for Systemic Reform, in: Bazemore/Walgrave (ed.), Restorative Juvenile Justice: in: Repairing the Harm of Youth Crime，1999，pp.45-74.第一個使用“最大模式”用以與“純粹模式”區別的學者正是瓦[19] 格雷夫。參見Walgrave, What is at Stake in Restorative Justice for Juveniles, in :Walgrave(ed.) Restorative Justice for Juveniles, supra, pp.11-16.丹尼爾•W•凡奈斯.全球視野下的恢復性司法//狄小華、李志剛.刑[20] 事司法前沿問題——恢復性司法研究，群眾出版社，2005：67-70.溫景雄.恢復性司法比較研究//狄小華、李志剛.刑事司法前沿問[21] 題——恢復性司法研究，北京：群眾出版社,2005:132.托克維爾.論美國的民主.董果良.北京：商務印書館,1997:67、[22] 282、283、284、290、291、299、301、303、314.國內許多學者認為恢復性司法予刑事和解的含義完全相同。黎[23] 宏.刑事和解：一種對傳統刑法理念的挑戰//黃京平,甄貞.和諧社會語境下的刑事和解,北京：清華大學出版社，2006:362。但是，在另一些學者看來，二者必須區分，恢復性司法的理論外延更為廣闊。如陳光中,葛琳.刑事和解初探.中國法學,2006(5):4.陳瑞華.刑事訴訟的私力合作模式—刑事和解在中國的興起.中國[24] 法學.2006(5):20.陳光中,葛琳.刑事和解的理論依據與適用構想//黃京平,甄貞.和諧[25] 社會語境下的刑事和解.北京：清華大學出版社，2006：21.郭晶.論構建刑事和解制度的訴訟階段間銜接機制—以探討檢察[26] 機關職權定位為線索.法制與社會,2009(3).杜宇.“刑事和解”的概念框架.中山大學法律評論[M].北京：法[27] 律出版社，2010:327.Walgrave, How Pure Can a Maximalist Approach to Restorative [28] Justice Remain? Can a Purist Model of Restorative justice Become Maximalist，supra,p.419.卞建林.美國刑事訴訟規則和證據規則.北京：中國政法大學出版[29] 社，1996:10.國內早期關於刑事和解的論著大多採用這種路徑。如劉淩梅.西[30] 方刑事和解理論與實踐介評.現代法學,2001(1).；吳宗憲.恢復性司法述評.江蘇公安高等專科學校學報.2002(3)；劉方權,陳曉雲.西方刑事和解理論基礎介評.雲南法學,2003(1)等.採用這一路徑的研究成果較多，馬靜華.刑事和解的理論基礎及[31] 其在我國的制度建構.法律科學,2003(4)；劉方權.恢復性司法——一個概念性框架.山東員警學院學報,2005(1)；邵軍.恢復性司法的利弊之爭.法學,2005(5)；向朝陽,馬靜華.刑事和解的價值構造及中國模式的構建.中國法學,2003(6)；宋英輝,許身健.恢復性司法程式之思考.現代法學,2004(3)；王俊.恢復性司法及其本土化改造.人民檢察,2006(2)；陳光中.刑事和解的理論基礎與司法適用.人民檢察,2006（5）；吳常青.論恢復性司法的本土資源與制
