運用隱喻計算於特色結盟之企業夥伴推薦研究 - 以區域觀光產業為例 - 政大學術集成

全文

(1)國立政治大學資訊管理學系. 碩士學位論文指導教授:苑守慈博士. 政治大運用隱喻計算於特色結盟之企業夥伴推薦研究立 ‧. ‧ 國. 學. - 以區域觀光產業為例. sit. y. Nat. Metaphor-Based Alliance Partners Recommendation for. n. al. er. io. Unique and Attractive Destination Image Building. Ch. engchi. i n U. 研究生：葉又誠中華民國 100 年 6 月. v.

(2) 致謝終於撐到寫致謝了，這段日子實在是……，唉!我說不下去了。總之，我必須感謝很多人，最重要的，當然是苑老師。苑老師是我在大學以來見過對學生付出最多的老師，不論是論文上的細心指導、英文上聽說讀寫的訓練、甚至設法支援有經濟困難的我，受到老師幫助的事項真的多不勝數，因此，我在心底給苑老師深深的一鞠躬，感謝她在這兩年來對我的栽培與協助。我感謝同實驗室的夥伴們。我感謝 Aflred 與 Mika 學長，在研究上他們著實. 治政大我們一起解決同樣的困難與問題，有她的陪伴，讓我感覺我不是一個人孤軍奮立. 了給我很多寶貴的建議。特別感謝淳雅，她大概我這兩年來最密集互動的夥伴，. 戰。我感謝 Claude、Claire、Sherry、Diana 跟 Kenny，雖然我們只有相處一年，. ‧ 國. 學. 但你們的加入讓我們實驗室增加許多活力與歡笑，讓我苦悶的研究生活多了許多. ‧. 樂趣。. y. Nat. 此外，我要感謝我的家人，在這兩年中對我無微不至的照顧，讓我得以專心. er. io. sit. 在學業上奮鬥。也要感謝提供我兼職工作的老闆，由於他提供我的這份工作，我才得以用最彈性的方式紓解我的經濟困難。最後，我要感謝其他所有在研究所生. al. n. v i n 活中認識的老師、同學與學長姐們，有你們的陪伴使得我的研究所生活更佳充實 Ch engchi U 與快樂。. 研究所大概是我最後一段學生生活，我感謝在家人們以及我成長過程中所遇到所有的朋友們跟老師們，沒有你們，沒有今天的我。. 葉又誠於台北家中 2011 年 7 月. II.

(3) 關鍵詞: 中小型企業、地點意象建立、隱喻計算、結盟夥伴選擇. 論文提要內容: 對於結盟的建立而言，如何選擇夥伴是相當重要的議題。許多的學術研究著重於建立一些選擇夥伴的框架或準則，以求達到資源分享、節省成本的效果。在旅遊產業中，許多文獻舉出了意象建立的重要性，也點出了意象的有效建立有賴於企業體彼此緊密的合作，然而，較少研究探討如果要建立獨特且具有吸引力的意象效果，應該選擇那些夥伴才能到到目標。因此，本研究提出一系統化的方法. 治政大一方法利用隱喻計算作為工具，嘗試找出創新的解決方案。本研究提出也提出一立. 能幫助使用者分析並找出合適的合作夥伴，以建立獨特且具有吸引力的意象。此. 個系統架構，並輔以相關的演算法與情境來說明方法上的可用性。從理論上的觀. ‧ 國. 學. 點來看，本研究嘗試透過自動化的方式找出隱喻的意涵，並將之整合到一問題解. ‧. 決的方法上。從實務面來看，本研究提供了中小型企業一個有用的方法能幫助他. y. Nat. 們找到合適的合作夥伴。透過建立更高品質的夥伴關係，我們期盼在旅遊產業的. er. io. sit. 中小型企業能夠進一步增加其競爭優勢、存活與獲利能力。此外，研究也發現，一個區域的意象多樣性直接影響到中小型企業透過合作來建立市場利基的可能. n. al. 性。. Ch. engchi. III. i n U. v.

(4) Abstract Partner selection is an important issue in alliance formation. A lot of research works have been done in developing the framework or criteria for selecting partners from the views of resource complement, cost reductions and knowledge sharing. However, research to date suggests relatively little is known about how to select partners for attractive and unique image building, which is essential to the developments of tourism especially for SME owners in the tourism sector. In this. 政治大 appropriate partners to form alliances and build their attractive and unique images. 立. paper, we propose a systematic approach for service providers in tourism to identify. This approach employs metaphors as a tool to generate innovative and creative. ‧ 國. 學. solutions. The system architecture is then provided and elaborated with algorithms. ‧. and the system scenario. From the theoretical perspective, we attempt to excavate the. sit. y. Nat. meaning of metaphors from the web in order to propose a new frame of. io. er. problem-solving. From the practical perspective, we provide SME owners with a useful approach for managing partner selection and attractive and unique image. al. n. v i n C h SMEs in tourism building. By forming better alliances, sector can gain competitive engchi U advantages and improve their sustainability and profitability. In addition, the image. diversity of a tourism destination is an important factor on market niche creation through alliance formation.. Keywords: SMEs, destination image building, computing metaphor, alliance partner selection. IV.

(5) TABLE OF CONTENTS. CHAPTER 1. INTRODUCTION .................................................................................. 1 1.1 Research Background ...................................................................................... 1 1.2 Research Motivations and Questions............................................................... 2 1.3 Research Objectives ......................................................................................... 4 1.4 Research Method ............................................................................................. 5 1.5 Research Contributions .................................................................................... 6 1.6 Content Organization ....................................................................................... 7 CHAPTER 2. LITERATURE REVIEW........................................................................ 9 2.1 Alliance Partner Selection ................................................................................ 9 2.2 Destination Image .......................................................................................... 15 2.3 Metaphor ........................................................................................................ 18 CHAPTER 3. MOTIVATION APPLICATION ........................................................... 23 3.1 The uVoyage Conceptual Framework ............................................................ 23 3.2 uVoyage System Architecture ........................................................................ 26 CHAPTER4. METAPHOR-BASED ALLIANCE PRATNER RECOMMENDATION MECHANISM ............................................................................................................. 29 4.1 Conceptual Framework .................................................................................. 29 4.2 Image Model .................................................................................................. 32 4.3 The System Architecture ................................................................................ 35 4.4 Goal Comprehension Module ........................................................................ 37 4.5 Candidates Generation Module...................................................................... 42 4.6 Niche Assessment Module ............................................................................. 49 4.7 Image Classification Module ......................................................................... 58 CHAPTER 5. Application Scenario ............................................................................. 61 5.1 An overview of application context ............................................................... 61 5.2 The service journey of the application ........................................................... 62 Chapter 6 Evaluation.................................................................................................... 68 6.1 Hypotheses ..................................................................................................... 68 6.2 Assumptions and Experimental Data............................................................. 70 6.3 Experiments and Results................................................................................ 75 6.4 Discussion of Findings .................................................................................. 92 Chapter 7 Conclusion ................................................................................................... 95 7.1 Contributions.................................................................................................. 95 7.2 Managerial Implications ................................................................................ 97. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. V. i n U. v.

(6) 7.4 Conclusion Remarks .................................................................................... 100. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. VI. i n U. v.

(7) LIST OF FIGURES Figure 1.1 Research framework (Source: Hevner et al, 2004)....................................... 8 Figure 2.1. The dimensions of destination image ........................................................ 16 Figure 3.1. uVoyage conceptual framework ................................................................ 24 Figure 3.2 uVoyage Service System Platform Architecture ......................................... 27 Figure 4.1 The conceptual framework ......................................................................... 30 Figure 4.2 Single Color Image Scale (Kobayashi, 1992) ............................................ 34 Figure 4.3 Key Word Image Scale (Kobayashi, 1992) ................................................ 34 Figure 4.4. System architecture ................................................................................... 36 Figure 4.5. Metaphor comprehension algorithm.......................................................... 40 Figure 4.6. Gap identification algorithm...................................................................... 42 Figure 4.7. Metaphor generation algorithm ................................................................. 44 Figure 4.8 the formula of image model similarity score .............................................. 46 Figure 4.9. Candidate Discovery algorithm (1) ........................................................... 47 Figure 4.10. Candidate Discovery algorithm (2) ......................................................... 48 Figure 4.11 The formula of goal fulfilment score ........................................................ 49 Figure 4.12. Goal fullfillment algorithm...................................................................... 50 Figure 4.13 The algorithm of attractiveness analysis component ................................ 53 Figure 4.14 The algorithm of uniqueness analysis component .................................... 56 Figure 5.1 The Service journey of business and customer .......................................... 63 Figure 5.2 The business registering form..................................................................... 64 Figure 5.3 The questionnaire for initializing business image model ........................... 64 Figure 5.4 the interface of metaphor-based partner recommendation system ............. 65 Figure 6.1. Market Niche coefficient formula ............................................................. 70 Figure 6.2 concentrated centers of image clusters (lower image diversity)................. 74 Figure 6.3 scattered centers of image clusters (higher image diversity)...................... 74 Figure 6.4.gap image coverage index formula............................................................. 76 Figure 6.5 The test run for deciding the number of metaphors required to ensure acceptable system efficiency performance .................................................................. 77 Figure 6.6.image discovery rate formula ..................................................................... 78 Figure 6.7 The number of Google queries ................................................................... 78 Figure 6.8 The similarity threshold for attractiveness analysis ................................... 81 Figure 6.9 The level of goal fulfillment under low level image diversity context ...... 83 Figure 6.10 The level of goal fulfillment under high level image diversity context ... 84 Figure 6.11 The level of improvement on goal fulfillment for each goal in two different context settings............................................................................................................. 84. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. VII. i n U. v.

(8) Figure 6.12 The level of uniqueness under different level of image diversity setting . 87 Figure 6.13 The level of uniqueness improvement under different level of image diversity setting ............................................................................................................ 88 Figure 6.14 The level of attractiveness improvement under different level of image diversity setting ............................................................................................................ 90 Figure 6.15 Market niche coefficient ........................................................................... 92. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. VIII. i n U. v.

(9) LIST OF TABLES Table 2.1 Criteria and sub-criteria for the partner selection (source: Wu et. al. 2009) 11 Table 2.2 Methods for partner selection .................................................................... 12 Table 4.1 An example of destination image model ...................................................... 33 Table 3.2 An example of gap image analysis ............................................................... 41 Table 4.3 An example of image model similarity analysis .......................................... 46 Table 4.4 An example of two image models ................................................................ 54 Table 4.5 The reference points for attractiveness and uniqueness score...................... 57 Table 5.1 An example of metaphor-based partner recommendation process............... 66 Table 6.1. The parameter setting for the experiments .................................................. 81 Table 6.2. The testing goalss prepared for the experiments ......................................... 82. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. IX. i n U. v.

(10) CHAPTER 1. INTRODUCTION Image building serves as a significant positioning strategy in the tourism industry. The research conducted for this dissertation uses computing metaphor to lay emphasis on developing a systematic approach for forming alliances in order to build attractive and unique images. This chapter introduces the core topics addressed in this dissertation: alliance partner selection, image building and computing metaphors. It offers an overview of interrelationships between these concepts and elucidates research background, motivations, questions, method and contributions.. 立. 1.1 Research Background. 政治大. ‧ 國. 學. In tourism industry, Small- and Medium-sized enterprises (SMEs) are those actors who account for a significant proportion of tourism service providers. For example,. ‧. 99% of tourism achievements in Austria are constituted by SMEs (OECD 2008).. sit. y. Nat. Tourism SMEs usually have poor resources for human resource development, limited. io. er. budget for marketing and lack of knowledge in recognizing their important roles in. al. v i n C asking “what can we do for them tohmake i U sustainable?” e nthem g c beh more n. tourism development. Due to these facts, researchers and governments still keep. In an effort to answer this question, the research suggests that SMEs in tourism to join a co-operative scheme in order to increase their performance and profitability. Tourists usually expect to attain a holistic experience from a destination, but such an experience often cannot be satisfied by a single small business service provider (OECD 2008). Therefore, there have been numerous studies in the literature about the prominence of cooperation and partnerships in the tourism sector (OECD 2008; Smeral 1998; Reid 2008). The report from Organization for Economic Co-operation and Development (OECD) indicates the success of an individual business often 1.

(11) depends on the success of a destination which can be derived from greater cooperation between tourism SMEs in the specific context of local networks and clusters (OECD 2008). Through cooperating with others, SMEs are capable to diversify tourism product portfolios to attract tourists and, in turn, improve their sustainability and profitability by making it possible for tourists to stay longer and consume more. In addition, it has been widely recognized that it is important to build unique and attractive destination images for tourism development (e.g., Mackay & Fesenmaier 1997; Vanolo 2004). Destination image might be thought of as “an impression of a. 政治大 the more chances tourists select the destination. Thus, positive, appealing and 立. destination (Echtner& Ritchie 2003).” The more attractive a destination image is built,. charming image-building of a destination can serve as a policy and strategic tool in. ‧ 國. 學. order to attract tourists, economic actors and investments (Vanolo 2004).. ‧. However, a gap has been identified between proposed destination images and. sit. y. Nat. supplied tourism products (Camprub ı´ 2008) because destination images are too. io. er. extensive to be built or provided by only one single SME business. In order to reduce this gap, cooperation between businesses is needed. In the context of all of the above,. al. n. v i n C htourism development the key means to further ongoing in a regional area will be to engchi U form highly integrated destinations with flexible operating network alliance. In. alliance formation process, partner selection is undoubtedly an important task for future success (Medcof .1997).. 1.2 Research Motivations and Questions A number of literatures have investigated how to select appropriate partners in order to form a prosperous alliance. Brouthers proposed a framework for analyzing the likely success of strategic alliances, called ‘the 4 Cs’, which involves complementary 2.

(12) skills, cooperative cultures, compatible goals, commensurate levels of risk (Brouthers et al. 1995). Several studies (e.g., Pesa¨maa et al. 2007; Shah et al. 2008) have also identified a series of criteria (e.g., trust, loyalty, complementarity, financial payoff) for selecting partners. While considerable attention has been paid in the past to research issues related to develop the partner selection criteria for the intentions of resource complement (Lin et al. 2009; Shah et al. 2008), cost reductions (Geringer 1991; Lin et al. 2009; Medcof 1997) and knowledge sharing (Brouthers et al. 1995; Dacin et al. 1997), there is no. 政治大 attractive and unique features (e.g., image building in tourism). 立. work on establishing a systematic approach to identify partner portfolios with the. ‧ 國. 學. In fact, partner selection is inherently time-consuming process. When practitioners deal with this kind of potential partnership discovering process, it’s. ‧. burdensome for them to identify every possibility of partner compositions.. sit. y. Nat. Nevertheless, very few studies attempt to develop an information system to facilitate. io. er. this process in an automatic way.. al. v i n C service-dominant logic (Vargo & hLusch 2008). Iti involves pre-visit and post-visit eng ch U n. For image building, it is inherently a value co-creation process as indicated in the. stages. When a tourist visits a tourism destination, he/she would have intensive interactions with the destination. The image of a destination is then co-created both by tourists and service providers within the destination. However, past research in image building often reflects a goods-dominant orientation of value creation (Kotler et al. 1999; Yu¨ ksel & Akgu¨ l 2007; Mackay & Fesenmaier 1997). They viewed the customers as operand resources and focus on making use of different marketing approaches to influence and change how the customers perceive the image of a destination in their image building processes. This 3.

(13) perspective somewhat limits the value co-creation opportunities. Therefore, we would like to develop an approach in the form of an information system enacts as an operant resource which can be leveraged by the SME owners co-creating the value with tourists (i.e., the other operant in destination tourism) in order to identify appropriate partners so as to integrate resources more effectively and propose the compelling value propositions (i.e., destination images). In sum, we recognized the importance of cooperation between SMEs in tourism industry and briefly reviewed partner selection criteria. We also noticed that image. 政治大 cooperation. However, we finally discovered that there is a valuable research question 立 building play a critical role in tourism success and its interdependency with business. ‧ 國. 學. still remains unanswered. That is, how do we help SMEs identify their alliance partners in order to build attractive and unique images? Is there any mechanism that. ‧. can be designed to aid SME owners to choose partners for image building? To what. sit. y. Nat. extent is the proposed mechanism leveraged to establish market niche for the SME. io. n. al. er. owners? These questions then lead to the following research objectives.. 1.3 Research Objectives. Ch. engchi. i n U. v. The overall goal of this research is to develop a creative and automatic approach that can facilitate partner selection process in order to build attractive and unique images. Through erecting differential images and forming novel partnerships, new niche markets and new tourism products could be explored and developed. To achieve this goal, the following list of more specific objectives is established. (1) To identify a way for innovative partner composition discovery. (2) To develop a system architecture and a series of algorithms that forms a basis for facilitating this alliance partner recommendation service. 4.

(14) (3) To implement a prototype system to demonstrate the feasibility and practicability of the proposed method.. By successfully forming a unique and attractive image, SMEs can seek to build promising alliances together, and thus improve their sustainability and profitability. 1.4 Research Method In this study, the proposed approach employs metaphor as a starting point. Metaphor is a structure of our cognitive system (Lakoff 1987) and affects the way we perceive. 政治大. the world, categorize experiences, and organize our thoughts (Casakin 2007).For. 立. example, the metaphor “my lawyer is a shark” entails that the lawyer is aggressive or. ‧ 國. 學. terrifying. A metaphor usually highlights certain aspects of the concept (i.e. aggressive) and downplays others (i.e. a creature). One of the utilities of metaphors is to help. ‧. Nat. sit. are thus a great form to represent abstract concepts like images.. y. people to define abstract concepts into a more concrete level (Hill 1995). Metaphors. n. al. er. io. Metaphors have been used to foster creative and innovative thinking in different. i n U. v. domains. For example, metaphors were applied to product design and architectural. Ch. engchi. design (Casakin 2007; Wang 2009). In product design, metaphors are used to explore the possibilities of product design solutions, and the product designers make their design to reflect the characteristics of metaphors to products on visual level, action level and image level (Wang & Liao 2009). Additionally, researcher and practitioners often make analogy between business environments and ecosystem so as to develop a series of strategies (Iansiti & Levien 2004). We have seen evidences of the usefulness of metaphors to reinforce innovative thinking. Therefore, we believe it can be applied to solve our research questions. Furthermore, computing metaphor is one of interesting topics in natural language 5.

(15) processing domain. Fruitful researches have produced lots of techniques to excavate the meanings behind metaphors (Abe & Nakagawa 2006; Baumer et. al. 2009; D'Harris 2002;Jones 1992; Mason 2004; Martin 1990; Slack 1980; Veale & Hao 2007; Weiner 1984; Zhou et al. 2007). These researches provide a solid basis for us to be able to develop a metaphor-based partner recommendation system. In general, the main reasons why we employ computing metaphors as the kernel of our approach are as follows. (1) Metaphors are great meaning carriers which help us to define abstract concepts. 政治大 (2) Metaphors have been widely recognized that its ability to foster unconventional 立 like images.. solutions discovering.. ‧ 國. 學. (3) Computing metaphor theory provides us a solid basis to analyze metaphors. ‧. automatically.. sit. y. Nat. io. er. In the scenario of using metaphor-based partner recommendation system, the users will be asked to provide a metaphorical statement which signifies the images. n. al. they want to build. Then. v i n theC system starts to analyze h e n g c h i U the. metaphor and finally. generate a series of partner candidates. Through integrating computing metaphors, we are able to develop a generative of discovering alliance partners for building the unique and attractive destination images.. 1.5 Research Contributions This dissertation presents the following key contributions: (1) From the theoretical perspective, we attempt to adopt computing metaphor to propose a new frame of problem-solving. (2) From the practical perspective, we provide SME owners with a useful approach 6.

(16) for managing partner selection and attractive and unique image building. (3) From service science perspective, we develop a systematic approach to facilitate value network development and service integration.. 1.6 Content Organization The research framework of this study is presented in Figure 1.1. In this chapter, the environment where the problem space resides is briefly described in terms of people, organizations and technology. SME owners and tourists are two kinds of people who. 政治大 provide motivations for SMEs 立. play active roles in this research. Businesses usually have scarce resources and limited capabilities. Such calls. into how to leverage. information technology to improve their sustainability and profitability. In order to. ‧ 國. 學. finely investigate and design innovative artifacts to tackle this challenging problem,. ‧. related literatures addressing alliance partner selection, destination image and. sit. y. Nat. metaphors will be provided in chapter two. The chapter serves as the fundamental. io. er. knowledge for us to further extend the human and organizational capabilities by creating new artifacts. Chapter 3 then delineates the whole picture of our research. al. n. v i n work and demonstrates the role C of the proposed mechanism h e n g c h i U in the application context.. Next, Chapter 4 elaborates our metaphor-based alliance partner recommendation approach including the conceptual framework that guides the system design and the technical detail of system architecture. For the purpose of validating the effectiveness and quality of the new artifact, Chapter 5 provides a detailed scenario and Chapter 6 offers a set of experiments to manifest its utility and performance. Finally, the research findings, implications and future research are addressed in Chapter 7 for researchers and practitioners to extend and implement.. 7.

(17) 立. 政治大. ‧. ‧ 國. 學. Figure 1.1 Research framework (Source: Hevner et al, 2004). n. er. io. sit. y. Nat. al. Ch. engchi. 8. i n U. v.

(18) CHAPTER 2. LITERATURE REVIEW. Chapter 2 reviews relevant research from alliance partner selection, destination image and metaphors. First, justification for a growing need of a systematic approach in partner recommendation is offered by examining current research on alliance partner selection. The literature regarding this issue is discussed to provide (1) a review of partner selection criteria and (2) a primary justification for establishing a creative method to excavate innovative partner composition. Following this discussion, a. 政治大. review of the destination image literature is provided to justify the importance of. 立. image building in tourism sector and amplify the interrelationship between image. ‧ 國. 學. building and alliance formation. Finally, the introduction of metaphors is given to justify the application of computing metaphor as a sound tool to facilitate the. ‧. innovative partner selection process.. y. Nat. sit. Chapter 2 has three objectives and it serves as the theoretical foundation. n. al. er. io. specified in research framework in Figure 1.1. The first is to review common. i n U. v. approaches used for alliance partner selection. The second is to provide theoretical. Ch. engchi. justification for sustainability improvements by the aid of image building and cooperation. The third is to provide the evidence that metaphors have great potential for tackling partner recommendation problem with the purpose of attractive and unique image building.. 2.1 Alliance Partner Selection Alliance can provide firms with new source of competitive advantages (Bierly and Gallagher, 2007). A number of studies have sought to identify the underlying 9.

(19) motivations for the alliance formation. These motivations majorly include resource complement (Lin et al. 2009; Shah et al. 2008), transaction cost reductions (Geringer 1991; Lin et al. 2009; Medcof 1997) and knowledge sharing (Brouthers et al. 1995; Dacin et al. 1997). In the resource complement perspective, it is innately superior to have partners with different resources that can provide absent ingredients or capabilities so as to leverage and integrate them to create synergies (Lin et al. 2009; Shah et al. 2008). On the other hand, transaction cost can be an important concern (Geringer 1991; Lin et al. 2009; Medcof 1997). Organizations tend to seek the. 政治大 efforts and money. Meanwhile, knowledge sharing can be another consideration. 立 partners who can reduce their business transactions in order to do more with less. Learning through cooperation could be one of the efficient and effective ways to. ‧ 國. 學. gather additional expertise and skills of specific areas (Brouthers et al. 1995; Dacin et. ‧. al. 1997). However, despite the growing numbers and increasing significance of. y. sit. io. er. Ross, J 2001).. Nat. alliances, considerable proportions of alliances performed ineffectively (Inkpen &. al. v i n Ch U management (Holmberg causes are inappropriate partner selection e n gandcpoor h i alliance n. The reasons behind the ineffectiveness of alliance are complex. Two common. & Cummings 2009). In this study, we focus on partner selection. The rich body of literature thus have explored and developed the partner selection approaches, checklists and criteria for partner selection (Brouthers et al. 1995; Dacin et al. 1997; Geringer 1991; Lin et al. 2009; Medcof 1997;Shah et al. 2008; Wu et. al. 2009). The paper in Wu (2009) provided a review of the partner selection criteria developed by numerous studies (see Table 2.1). The common criteria include characteristics of the partner,. marketing. knowledge. capability,. 10. intangible. assets,. complimentary.

(20) capabilities and degree of fitness. These criteria can be further subdivided into minor aspects for evaluating the fitness of partners.. Table 2.1 Criteria and sub-criteria for the partner selection (source: Wu et. al. 2009) Criteria. Sub-criteria. Characteristics of the Unique competencies, compatible management styles, partner. compatible strategic objectives, higher or equal level of technical. capabilities. between. manufacturers. 政治大. and. distributors. 立. knowledge of local business practices Trademarks,. patents,. licenses,. or. ‧. Intangible assets. 學. capability. ‧ 國. Marketing knowledge Increased market share, better export opportunities, and. other. proprietary. sit. y. Nat. knowledge, reputation, previous alliance experiences,. io capabilities. al. n. Complimentary. er. technically skilled employees among partners. i n U. v. Partners owned managerial capabilities, wider market. Ch. engchi. coverage, diverse customer, the quality of distribution system to those of the strategic partners. Degree of fitness. Compatible organization cultures, willingness to share expertise, equivalent of control. Such criteria contributes to the research works for developing systematic methods in partner selection. Above-listed criteria are often integrated into the development of partner selection methods. Table 2.2 presents the methods for partner 11.

(21) selection and their criteria used. In fact, most methods are concentrating functional aspects (i.e., cost, quality, performance, etc) of partner selection, especially those used in developing production network or supply chain management (Amid et. al. 2006;Chang 2006; Feng et. al. 2010; Fischer et. al. 2004; Hacklin et. al. 2006; Jung 2010; Yeh & Chuang 2010). These methods also demonstrate product-centric perspectives, which. regard products as the starting point of planning process and focus on performance, efficiency and utility (Sheth et. al. 2000). However, we argue that customer-centric perspective should be bought into. 政治大 involves cross-industries relationships resulting in alliances because tourists need 立. partner selection issue, particularly for service industry like tourism, which inherently. ‧ 國. 學. various kinds of service (i.e., transportation, accommodation, entertainment and so on) during a journey. To best practice the customer-centric perspective, the intrinsic value. ‧. (i.e., psychological or emotional value) should not be ignored. As observed in Table. sit. y. Nat. 2.2, these psychological and emotional aspects are also neglected in current methods. io. al. n. superior service experiences remains unanswered.. Ch. engchi. er. for partner selection. The problem regarding how to select partners in order to create. i n U. v. Table 2.2 Methods for partner selection. Authors. Methods. Criteria used. Target. (Jung 2010). Artificial Intelligence (fuzzy, AHP). product quality, performance history, reputation, production capacity. Production network. 12.

(22) rapid market entry, compatible management styles, political advantage, compatible strategic objectives, distribution network (Hajidimit Linear quality, willingness to share riou and programming( g expertise, compatible organization Georgiou oal programming cultures, better export opportunities, 2002) model) technological level, quality of local personnel, knowledge of local business practice, location of joint venture facilities. International Joint Venture. individual utility(technology capability, financial health, knowledge and managerial Fuzzy multiple experience, capability to access new attribute (Feng et. market ), collaborative utility decision-making al. 2010) (Resource complementarity, (FMADM) overlapping knowledge approach bases ,motivation correspondence, goal correspondence, compatible cultures). New product development. 立. 政治大. ‧. ‧ 國. 學. n. al. sit. er. io. (Ye 2010). y. Nat. TOPSIS (Technique for Order Preference cost, time, trust, risk, quality by Similarity to an Ideal Solution ). Ch. engchi U. v ni. Virtual enterprise. (Wang & Fuzzy preference price, quality, financial stability, Chen programming customer service, 2007) (FPP) method. Virtual Enterprise. (Chang 2006). Supply Chain. Fuzzy linguistic R&D, cost, quality, service, quantifier response. Fuzzy (Amid et. multiobjective al. 2006) linear model. cost, quality. service, capacity. (Fischer Ant colony number of offer, date of delivery, et. al. optimization price, social competence 2004) algorithms, AHP 13. Supply Chain. Production Network.

(23) ( Hacklin Decision support et. al. strategy, culture, structure system 2006). (Ding&Li ang 2005). Fuzzy. (Yeh & Multi-objective Chuang genetic 2010) algorithms. 立. Production Network. complementary capabilities, deeper contents and forms of collaboration, similarities match with partners, financial health, physical facilities and equipment, Intangible assets, market knowledge access. Liner shipping. green principles, production cost. production time, transportation cost, transportation time ,product quality. Supply chain management. 政治大. On the other hand, the research indicates that partner selection is time-consuming. ‧ 國. 學. process and the decisions of alliance formations are often made in the limited time and information (Holmberg & Cummings 2009; Bierly and Gallagher, 2007). For. ‧. tourism SMEs, the business owners rely on the knowledge of their social network to. y. Nat. sit. form the alliances. The search base of the list often is limited to those companies they. n. al. er. io. have known, and the opportunities for innovation consequently are constrained. Such. i n U. v. calls provide motivation for us to develop an information rich platform with a set of. Ch. engchi. semi-automatic and manageable mechanisms that can enlarge the potential partner base and facilitate the process. Therefore, for recent study, researchers have produced conceptual partner selection framework, checklists criteria and methods for partner selection, but little is known how to select partners in order to bring emotional value to their customers. To further facilitate partner selection process, our study intends to develop an alternative approach characterized by the ability to search in a wider candidate base automatically and the ability to evaluate partner portfolios while considering emotional value brought to customers. 14.

(24) 2.2 Destination Image In past decades, considerable concern has arisen over destination image building in the field of tourism research (Smeral 1998; Mackay & Fesenmaier 1997). This trend is basically derived from the fact that destination image has profound influence on tourists’ decision making process so that it can serve as a source of differentiation. Destination image, by definition, can be regard as the perceptions, beliefs, impressions, ideas and understandings that a person has of a destination (Tuohino 2001; Echtner& Ritchie 2003).That is, destinations with attractive, unique, compelling,. 政治大 & Lysonski 1989). This. positive images have more chances to be selected by the tourists during travel. 立. decision process (Woodside. phenomena implies that. ‧ 國. 學. understanding the images embedded in the mind of tourists is highly beneficial to explain tourists behaviour and develop effective destination marketing strategy.. ‧. According to Echtner and Ritchie (2003), destination image can be interpreted in. Nat. sit. y. terms of three dimensions –attributes to holistic, functional characteristics to. n. al. er. io. psychological characteristics and common to unique – as indicated in Figure 2.1.. i n U. v. Destination image is intrinsically impressions of place. In attribute-based perspective,. Ch. engchi. these impressions may refer to different attributes of a destination such as climate, price level, infrastructure, friendliness and so on. Alternatively, destination image can be described in holistic view. For example, Paris is commonly regarded as one of most romantic cities in the world. Among those image attributes, some of them are closer to functional characteristics, associated with tangible aspects of the destination, such as natural attractions and others are closer to psychological characteristics, tied with intangible aspects of the destination, such as atmosphere. Moreover, some of destination images may be very unique or common in comparison to other destinations. Unique attributes somewhat have a tourism destination better positioned 15.

(25) to compete with other destinations. By identifying the characteristics of a destination, tourism development agencies are able to manage and propose a compelling image to make a destination more sustainable. Above all, this framework pinpoints three dimensions that are genuinely useful for capturing the tourism-related elements which can serve as a starting point for considering tourism development.. 立. 政治大. ‧. ‧ 國. 學. n. al. er. io. sit. y. Nat. Figure 2.1. The dimensions of destination image. i n U. v. Recent research has shown the significance of marketing alliance for image. Ch. engchi. building and tourism development. This concern mainly derived from the argument that it is always possible to find the gap between the destination image the marketers are trying to make and supplied tourist products (Camprub ı´ 2008). That is, the proposed destination image may not reflect the reality of a destination owing to the low degree of collaboration and cohesion between businesses which supply the tourism products. This gap may have substantially negative impact on customer satisfaction. One possible explanation regarding this issue is that some destination images are too extensive to be formed or provided by only one single business. For example, if a destination is promoted as a paradise resort, it seems unlikely to exhibit 16.

(26) this image only by the efforts of one business. To create this image, it is supposed to involve different actors (i.e. accommodation service provider, outdoor activities provider, government, tourism development agencies, etc) within this area to cooperate with each other for offering the holistic experience of a destination to tourists. SMEs are those actors with scarce resources and limited capacity, but for major participants in tourism industry, it would be innately superior to engage them in developing a high level of cohesion, cooperation and coordination network (OECD 2008), which can be achieved by the efforts of multiple alliance relationships within a. 政治大 marketing alliance for image building being identified in the literature, scant research 立. destination. The image gap thus can be reduced. In spite of the importance of. specifically proposed systematic approaches to identify the appropriate partner. ‧ 國. 學. compositions for image building.. ‧. In our study, we further extend the notion of destination image to business image.. sit. y. Nat. Attractive and unique business image building is assumed to be a differentiation. io. al. er. strategy in tourism context. In examining previous research regarding image. iv n U this image. n. measurement, it should be noted that no matter it is a product, store or business, there. Ch. engchi. is a certain image perceived by customers and. strongly influences. customer’s decision making process (Echtner & Ritchie 2003; MacInnis & Price 1987; Hampton et al. 1987; Jain & Etgar 1976; Stell & Fisk 1986). Hence, through creating positive images and forming successful alliances, SMEs are capable of launching more alluring tourism products and then improve their capacity and profitability.. 17.

(27) 2.3 Metaphor Metaphors might be thought of as “understanding and experiencing one kind of things in terms of another (Lakoff & Johnson 1980).” For example, we understand abstract ideas of friendship in terms of our experience in taking care of flowers. The metaphor, “Friendship (i.e., target) is a flower (i.e., vehicle),” entails the ideas that friendship is beautiful and vulnerable and it can grow and bloom. These attributes of flower can be applied to the target “friendship”. The ideas from the attributes of flower influence the way “friendship” is understood. Therefore, metaphors are broadly considered to be a. 政治大 and the source domain (Lakoff & Johnson 1980; Mason 2004). The target domain 立. conceptual mapping of properties between two knowledge domains, the target domain. ‧ 國. 學. provides dimensions for attribution, whereas the source domain (or vehicles) offers properties that may be applicable to the target (McGlone & Manfredi 2001).. ‧. A metaphor, as a result, is an effective manner for defining abstract concepts on. Nat. sit. y. more concrete level. The abstract nature of friendship is made clearer by defining. n. al. er. io. more concrete characteristics of flower like beauty and vulnerability. Destination. i n U. v. image, in essence, is an abstract concept, which is suitable to be described in the form. Ch. engchi. of metaphor. By analyzing metaphors, the characteristics of abstract concepts are able to be embodied. Metaphors have been widely applied to many areas, such as computer science, psychology, the corporate world, and one of the most interesting applications is in design problem solving due to its potential on enhancing creative and innovative thinking (Casakin 2007; Lubart & Getz 1997; Weick 2003). When designers want to develop an innovative solution to a specific problem, the critical first step is to perceive the world in an unorthodox and unconventional way. Morgan once stated “metaphors provide some different ways of thinking about things (Morgan 2006).” 18.

(28) That means metaphors can help us uncover the complex and paradoxical characteristics of things, we then are able to manage and design the solutions that we may have not thought possible before. These ideas have appeared in the vast literature. For instance, in architectural domain, one of the most impressive metaphors ‘less is more’ makes reference to the engineering idea of reducing architectural design to its minimal and basic nature (Casakin 2007). In product design, metaphors are used to explore the possibilities of product design solutions, and the product designers make their design to reflect the. 政治大 (Wang & Liao 2009). In business administration, metaphors are regarded as a tool to 立 characteristics of metaphors to products on visual level, action level and image level. ‧ 國. 學. describe “visions” or organization mission and strategy to gain novel concepts for innovation (Hill & Levenhagen 1995). In our study, due to the capabilities of. ‧. metaphors in creative thinking, we believe it is promising to apply metaphors to. sit. y. Nat. design partner configurations for attractive and unique destination images building.. n. al. er. io. In the previous examples, most of the metaphors were interpreted and created by. i n U. v. human being. In most cases, people can understand the metaphor, but they are usually. Ch. engchi. unable to speak out all the specific meanings behind the metaphors (Zhou et al. 2007). Additionally, we often need to come up with an insightful metaphor before we really enjoy the advantages of metaphors. These are two important tasks we have to tackle when we are using the metaphors. As a result, a number of studies have investigated on how to support the above tasks in an automatic way, that is, computing metaphor. These studies can be generally classified into two categories. One is metaphor comprehension and the other is metaphor generation. Both tasks are designed to be done automatically by the aid of information technology. 19.

(29) Metaphor comprehension is defined as a process of mapping between the target and vehicle concepts in order to identify some similarities for metaphor interpretation (Slack 1980; Zhou et al. 2007). There are two general ways that dominate the metaphor comprehension task. One is the rule-based approach and the other is statistics-based approach (Zhou et al. 2007). Approaches based on rules are usually involving hand-coded rules or knowledge base (D'Harris 2002; Martin 1990; Weiner 1984). The level of applicability in these system is limited to the predefine knowledge base. Notice that it is rather difficult to define all the rules or knowledge by human. 政治大 dynamically mining documents or corpus on the fly to understand the metaphor 立. beings. Statistic-based approaches, alternatively, could be implemented by. components (Mason 2004; Veale & Hao 2007). The corpus from web serves as a. ‧ 國. 學. plentiful knowledge source that implicitly represents a different perspective in the. ‧. world.. sit. y. Nat. Similar to metaphor comprehension, metaphor generation process involves. io. er. identifying the vehicles associated with shared common attributes (Abe & Nakagawa. al. v i n C h However, there were a novel metaphor is more complex. e n g c h i U still some approaches, e.g. n. 2006). The approaches for metaphor generation are relatively few because generating. Sardonicus (Veale & Hao 2007) and transparently-motivated (T-M) (Jones 1992). These approaches also can be classified into two categories, statistic-based approaches and rule-based approaches. Statistic approaches are normally developed based on leveraging statistic method in mining the corpus to establish a probabilistic model or identify the concept patterns (Abe & Nakagawa 2006;Veale & Hao 2007). Rule-based approaches can be implemented as building knowledge base through framing grammatical or hierarchical structure relationships between target and source domains (Baumer et. al. 2009; Jones 1992). 20.

(30) Although a substantial body of research studies are now available to shed light on automatically understanding or generating metaphors, the application of computing metaphors is still at early stage. We can observe metaphors have been used in vast domain like mentioned above. We also can notice that, in natural language processing domain, there are paramount studies on processing metaphor automatically (Abe & Nakagawa 2006; Baumer et. al. 2009; D'Harris 2002; Jones 1992; Mason 2004; Martin 1990; Slack 1980; Veale & Hao 2007; Weiner 1984; Zhou et al. 2007). Nevertheless, scant research deals with integrating computing metaphor approaches to. 政治大 metaphors to explore innovative solutions for alliance partner selection in a heuristic 立. solve the problem in specific industries. In this study, we aim to take advantages of. and automatic manner in the regional tourism industry.. ‧. ‧ 國. 學. This chapter has reviewed literature from several issues – alliance partner. Nat. sit. y. selection, destination image and metaphor – relevant to the goals of this dissertation.. n. al. er. io. This review included discussions on traditional way of partner selection, influential. i n U. v. accounts of destination image building and the utility of metaphors. We can then. Ch. summarize three points as follows.. engchi. (1) For partner selection, there is growing need on a systematic and automatic approach to uncover partner compositions with market niche potentials. (2) For SMEs in tourism industry, it is beneficial for them to make successful alliances by building attractive and unique images. (3) For image building, metaphor is a great meaning carrier and its power on creativity contributes to innovative solutions discovery. Computing metaphor theory here forms a sound basis for analyzing metaphor automatically. 21.

(31) Through connecting these concepts and theory, we are able to develop a methodology to solve our research problem in an unconventional way.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 22. i n U. v.

(32) CHAPTER 3. MOTIVATION APPLICATION The purpose of this chapter is to give a brief introduction of our research project called “uVoyage” that led to the development of proposed approach. This chapter describes the design of uVoyage conceptual framework, the rationales behind various aspects of the design and the role of alliance partner recommendation approach in this framework. This chapter focuses solely on giving the big picture of our research project and it serves as the designed artifact specified in research framework in Figure 1.1. The following chapter presents the technological details of the proposed partner recommendation approach.. 學. ‧ 國. 立. 政治大. 3.1 The uVoyage Conceptual Framework. ‧. At the heart of uVoyage conceptual framework is to improve the collective health of the SMEs in tourism regions/clusters that can assist the creation and delivery of. y. Nat. er. io. sit. service products by a set of particularly designed mechanisms leveraging information and communication technology (ICT). Under this proposition, the main issues are. al. n. including as follows.. Ch. engchi. i n U. v. (1) For businesses, how do service providers share resources and capabilities in order to improve collective sustainability? (2) For customers, how do people discover the unknown but desired services matched with their emotional needs? (3) For tourism development, how do administrators develop highly cooperative value network and assess the health of regional tourism ecosystem?. 23.

(33) Destination Ecosystem Health Common Assets Diversity Metric. Third Party Services. Alliance Network Design & Assessment Computing Metaphor. Sheltering Operations. Image Modeling Destination. Technology. Relationship. Figure 3.1. uVoyage conceptual framework. 政治大. The uVoyage conceptual framework consists of four tiers as shown in Figure 3.1.. 立. At the bottom of the cube is the first tier indicating three basic elements of regional. ‧ 國. 學. tourism. Among them, destination refers to a certain tourism location and its environment that many tourism-related service providers live by. Within a destination,. ‧. These three basic elements set up an. sit. Nat. between B2B, B2C and C2C are also focused.. y. technology plays a significant role in facilitating tourism services. The relationships. er. io. ideological foundation for us to consider region/cluster tourism development.. al. n. v i n image C modeling and sheltering h e n g c h i U operations. In the second tier,. enable service. providers to manage their images, and control risks and unwanted fluctuation in current or future business. Image modeling is an automated and adaptive mechanism for modeling the image features of destinations, customers and businesses. As described in chapter 2, destination and business images reflect the thoughts or feelings of customers towards a destination or business. Customer images, on the other hand, are modeled to reflect the personalities, preferences and emotional needs of a customer. Once these three kinds of images are modeled, we are able to manage images for businesses or destinations and match customers’ needs to desired businesses or destinations. In addition, sheltering operations provide a series of tools 24.

(34) that relieve certain marketing or managerial operations performed by businesses. By getting additional resources or capabilities, the performance or productivity of business can be further enhanced. In third tier, common assets and third party services are included. Based on image models in the second tier, we are able to perform alliance network design and assessment by integrating computing metaphor technique. A mechanism for evaluating service diversity of regional tourism ecosystem is also provided. Furthermore, Third party service providers, such as transportation, logistics, raw. 政治大 built on top of the common立 assets in order to facilitate or enrich the service operations. material suppliers, or other independent software vendors, enact as auxiliary enablers. ‧. ‧ 國. and improved.. 學. On the all the tiers mentioned above, the destination ecosystem heath can be measured. In general, uVoyage conceptual framework includes a set of systematic. y. Nat. sit. approaches that focus on fostering regional tourism development and improve its. n. al. er. io. sustainability. The core values of the framework are described in terms of the following 3P (Protective, Proactive and Prosperous).. Ch. engchi. i n U. v. Protective : Protective means that itis designed for SMEs in order to create a financial safety net. That is, protect SMEs against failures and then improve their survivability. Proactive : Proactive implies that SMEs can keep being alerted the changes of environments and customer needs based on the image modeling. SMEs then can proactively take actions to response the ever-changing environments. Prosperous : Prosperous denotes the ideal outcome of regional tourism service system. The underlying framework encourages SMEs to design better value 25.

(35) propositions to make an attractive change that contributes to making a destination more prosperous and improving collective sustainability.. With the proposed uVoyage conceptual framework , we will implement it in the form of web-based platform called “uVoyage”. The following section illustrates the system architecture and modules of uVoyage platform.. 政治大. 3.2 uVoyage System Architecture. 立. There are mainly six modules in uVoyage as shown in Figure 3.2. The modules of. ‧ 國. 學. image modeling and image mixing are related to the first two tiers of theframework.. ‧. These two modules attempt to capture destination, business and customer images in order to have a protective mechanism that senses the dynamic environment and. y. Nat. er. io. sit. customer needs. The sheltering service management module is designed for realizing the sheltering operations for SMEs mentioned above. Additionally, SME alliance. n. al. Ch. i n U. v. service formation and alliance feasibility management module are designed for. engchi. assisting business cooperation and value sharing. Finally, destination service module is developed to match tourists’ expectation and the tourism services available on the platform.. 26.

(36) Environment Data. Tourist Preference Input. SMEs Data. Image Modeling. Destination Service Matching. Alliance Feasibility Measurement. SME Alliance Service Formation. Image Mixing Sheltering Service Management. uVoyage B2C / B2B Service Modules. uVoyage Service Design. uVoyage Service Experiences. 政治大. Figure 3.2 uVoyage Service System Platform Architecture. 立. Both tourists and regional tourism SMEs are the target users of uVoyage. Tourists. ‧ 國. 學. search for desired services, whereas SMEs search for potential partners to cooperate.. ‧. In essence, the overall system process is to model image, recommend alliance partners,. sit. y. Nat. measure alliance feasibility, form alliances, match customer needs to desired service. io. er. and finally offer sheltering service. More specifically, the system begins from capturing images of destination, businesses and tourists. Tourists’ preferences,. al. n. v i n C hdata are used as the regional SMEs’ and environment primary input for the image engchi U modeling. Image mixing module here aims at making image model more. representative. We assume the interaction between tourists, SMEs and the environments will influence their images. Thus, the image mixing module is designed for reflecting these interactions to image modules. Besides, in order to reduce computation complexity, we use colors as the uniform representations of images. After images have been modeled, alliance partner recommendation can be performed. The mechanism of alliance partner recommendation is main focus of this study. It will be elaborated in the following chapter. Next, every cooperation 27.

(37) suggestion coming from the alliance service formation module will be evaluated by the alliance feasibility measurement module for assessing the possibility of cooperation success.. Afterward, the destination service matching module. recommends services to tourists by matching images of tourists to SMEs’ services. Finally, with the sheltering service management module that fulfills related electronic cooperation management and marketing functions, tourism SMEs can deliver their services more effectively. In summary, the uVoyageconceptual framework and the uVoyage service system. 政治大 The conceptual framework立 emphasizes the importance of value creation and value. design architecture was proposed for creating an innovative tourism service platform.. ‧ 國. 學. sharing with different stakeholders in the ecosystem. Regional tourism SMEs can create their attractive services through cooperation services to fulfill tourists’ needs. ‧. and obtain sheltering services from the uVoyage platform. Tourists can also more. sit. y. Nat. easily discover services which are more closely related to their thoughts/feeling and. n. al. er. io. compose the desired journey. Tourism SMEs on the uVoyage platform then keeps. i n U. v. getting feedback from tourists and find more potential partners to cooperate and. Ch. engchi. design more innovative services. Consequently, the sustainability of SMEs can be improved and the development of region/cluster tourism can also be fostered.. 28.

(38) CHAPTER4. METAPHOR-BASED ALLIANCE PRATNER RECOMMENDATION MECHANISM. This chapter describes the implementation details of metaphor-based alliance partner recommendation system and it serves as an artefact in terms of algorithms as specified in research framework in Figure 1.1. The primary purpose of this chapter is twofold. First, it elaborates the connections between different concepts including computing metaphor, image building and alliance partner selection and how to integrate them. 政治大. into the system architecture. Second, it offers a comprehensive description of system. 立. modules and exhibits its ability to provide a solution to manage partner selection and. ‧ 國. 學. attractive and unique image building. Throughout this chapter, we will provide several examples to demonstrate the required data and related computations in terms of. ‧. algorithms and formula performed at each step in the process so as to justify the. Nat. n. al. er. io. sit. y. feasibility and creativity of the system.. 4.1 Conceptual Framework. Ch. engchi. i n U. v. The underlying conceptual framework of this study is shown in Figure 4.1 prescribing the basic concept guiding our system design. The primary goal of our method is to identify possible alliance partner compositions for attractive and unique image building. The aim of this implementation, then, is to support this process by leveraging computing metaphor technique. By integrating different existing computing techniques, we further pursue a new metaphor-based frame for problem solving.. 29.

(39) As described in Chapter 2, image building building can serves as a positioning strategy directly contributing to the success of tourism destination and businesses. Image building, in essence, highly depends on the capabilities and resources possessed by business owners. To comprehensive image building, business business owners can acquire the missing elements from partnerships. Additional benefits coming from partnerships. include integrative tourism products portfolio development, information and knowledge sharing, and better customer service (Hamel et. al. 1989). With successful alliance formation, alluring images are likely to build and more tourism products are. 政治大 collaborative value network of a local area is built and thus drives this area being 立 able to be developed. Through this positive and iterative process, a highly. more successful and prosperous.. Computing Metaphor. ‧. ‧ 國. 學. Image. Metaphor. n. al. er. io. sit. y. Nat Attractive and Unique Image Building. Alliance Formation. Ch. engchi U. v n i Recommendation Partner. Local Tourism Development and Success. Figure 4.1 The conceptual framework. Nevertheless, there exists a huge gap between the image building and partner selection. How do business owners know which partners they should choose in order. to build a specific image? Is there any easier and systematic way that can help them 30.

(40) complete this job? Our study is intended to answer these questions. Metaphor here serves as a good starting point to reduce this gap owing to the following reasons. First, metaphors are genuinely useful in defining abstract concepts like images. In fact, destination images are often depicted in the form of metaphors (Vanolo 2004). For example, Hawaii is just like a paradise. Second, the notion of metaphor as an excellent facilitator for the design idea creation and innovative thinking is gaining credence. On the top of this, we believe metaphors have great potentials to discover unfamiliar solution to partner compositions problem. Third, available computing. 政治大 automatically completed. As mentioned above, the employment of metaphor is then 立 metaphor theory forms a theoretical grounding that the analysis of metaphors can be. ‧ 國. 學. justified.. This conceptual framework can be summarized by pointing out six. ‧. interrelationships that delineate the nature and contributions between concepts. These. Nat. sit. y. interrelationships consist of following:. n. al. er. io. (1) Metaphors are wonderful meaning carriers that help people understand abstract concept (e.g. images) on more concrete level.. Ch. engchi. i n U. v. (2) Computing metaphor technique forms a sound basis for us to analyze metaphors exploringly and automatically. (3) The true power behind metaphors is the ability to explore unfamiliar design alternatives and establish novel associations with the problem so that new possible partner compositions may be uncovered. (4) Through using our recommendation system, business owners are able to identify partners to form a featured alliance.. 31.

(41) (5) By forming alliance, the lacking capabilities or resources can be complemented to build attractive and unique images. (6) With highly development of local partnership network, the tourism destination will have more chances to succeed. The framework described in Figure. 4.1 provides a guide that demonstrates the big picture of this study. Next, a brief introduction of image model is provided. Based on the conceptual framework, the system architecture is developed and exhibited in the following subsections. However, prior to the presentation of our system architecture,. 政治大. the assumed modelling of images for destinations, businesses and customers will be. 立. briefed first.. ‧ 國. 學 ‧. 4.2 Image Model. sit. y. Nat. Image model is a model which is developed to capture emotional perceptions of a. io. er. destination or business. There are three kinds of image model in our system –. al. destination image model, business image model and customer image model. The. n. v i n meanings of destination image C model business image model are the same – the h eand ngchi U emotional perceptions of destination or business. One the other hand, the meaning of customer image model has a slight difference; that is, customer image model is used to describe their psychological preference or emotional needs towards businesses or destinations. The structure of an image model involves adjectives used to describe the target (a destination, business or customer) and an intensity value of each adjective. Intensity value is the percentage of people who think a specific adjective is appropriate to describe the target. Table 4.1 illustrates an example of destination 32.

(42) image model in which people think it’s charming, fascinating and enjoyable when they travel around this tourism site. The image element “enjoyable” with higher image intensity value imply this destination gives people relatively strong feeling of “enjoyable”. In our research, image model serves as a basis for us to further collect, present and analyze image data in a systematic way.. Table 4.1 An example of destination image model Image element No.. Adjective. RGB value. 1. charming. (255,18,204). 2. fascinating. of Image intensity. 政治 300大 (176,119,72) 200 (216,128,0). 0.3 0.2. 500. 0.5. 學. ‧ 國. 3. 立 enjoyable. Number people. In order to make image model be easily analyzed and processed, all of the image. ‧. elements (i.e., adjectives) can be represented by a color if necessary. A significant. y. Nat. sit. amount of research on color psychology has disclosed that colors are often associated. n. al. er. io. with feelings or emotions (Kobayashi, 1981; Nijdam, 2005; Ou et al., 2004; Suk and. i n U. v. Irtel, 2010; Xin et al., 1998). For example, the color red has been associated with. Ch. engchi. excitement and the color yellow has been associated with cheerfulness. Kobayashi provided an excellent model relating colors to emotions, called color image scale (see Figure 4.2 and 4.3). Through adopting this scale, every adjective manifesting the emotion perception of a target in image models can be associated a color with a RGB value. One of important considerations for mapping adjectives onto colors is that a quantitative method can analyze or compute those image data after the adjectives are replaced with RGB values. Under the constraint of mapping every word onto a color, the number of image elements in an image model is limited to 122 words, which can be found in Color Image Scale. 33.

(43) 立. 政治大. ‧ 國. 學 ‧. Figure 4.2 Single Color Image Scale (Kobayashi, 1992). n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Figure 4.3 Key Word Image Scale (Kobayashi, 1992) 34.

(44) 4.3 The System Architecture As mentioned before, the goal of this system is to leverage computing metaphors to excavate the meaning behind proposed images and further identify possible partner candidates. In order to help users to get desired outcomes, market niche assessment of different recommendation choices will be evaluated during the system process. To achieve these ends, the system architecture is designed and presented as Figure 4.4. In essence, the overall system process is to comprehend the goal, identify the missing image elements to achieve the goal, generate new metaphors, map these. 政治大. metaphors to a manageable set of candidates, and evaluate and prioritize the. 立. alternatives. Therefore, this system architecture consists of four main modules – goal. ‧ 國. 學. comprehension module, candidates generation module, niche assessment module and image classification module. As the names of modules imply, each module represents. ‧. the basic goal that the process needs to be completed in order to achieve the desired. Nat. sit. y. end. Before the details of these modules are provided, the structures of image model. n. al. er. io. should be formulated.. Ch. engchi. 35. i n U. v.

(45) 政治大. 立. ‧ 國. 學. Nat. y. ‧. Figure 4.4. System architecture. io. sit. (1) Goal comprehension module. n. al. er. Purpose: comprehend the meaning of the goal and identify the missing image elements for goal achievement.. Ch. engchi. i n U. v. Major Input: a goal in the form of a metaphorical statement Components: metaphor comprehension and gap identification Major Output: gap image elements (2) Candidates generation module: Purpose:conduct metaphor generation and comprehension process to generate a series of partner composition choices Major Input: gap image elements Components: metaphor generation, metaphor comprehension, candidate discovey and goal fulfilment analysis Major Output: a series of partner composition choices (3) Niche assessment module: 36.

(46) Purpose: evaluate the market niche potential for different choices Major Input: a series of partner composition choices Components: attractiveness analysis and uniqueness analysis Major Output: attractiveness score and unique score (4) Image classification module: Purpose: process image data in advance to reduce the computation complexity of niche assessment Major Input: customers’ image preferences and business images Components: customers’ preferences classification and business images classification. 政治大. Major Output: customer image preference clusters and business image clusters. 立. ‧. ‧ 國. 學. We will describe each of the modules in additional process details and provide the algorithm and formula of these modules.. 4.4 Goal Comprehension Module. y. Nat. sit. An assumption of our partner recommendation system is that the users have an ability. n. al. er. io. to clearly define their image they would like to build. For instance, a SME may want. i n U. v. to make its customers think they’re so happy and just like in paradise when they visit.. Ch. engchi. Then, a SME user needs to have an ability to think of the word “paradise” and input it into our system. After that, the system would start from analyzing the goal “paradise” and identify its latent meanings in order to find the most appropriate partners for collectively achieving the goal. The goal comprehension module is designed to perform this task. In order to comprehend metaphors, e.g., a SME (target) is just like a paradise (vehicle), we adopt a web-driven, case-based approach called the Sandonicus approach. (Veale & Hao 2007), which leverages the text of web as a plentiful. knowledge source to identify what properties are most contextually appropriate to 37.

(47) apply to both sides of target and vehicle. This approach employs Google search engine as a retrieval mechanism for finding properties of words by using Google supported APIs, which allow the search of wildcard term * as any possible words. For example, if you send a query “as * as paradise” to Google, you may get a series of words, such as beautiful, gorgeous, wonderful. That implies paradise can be beautiful, gorgeous and wonderful. More specifically, these words can be considered as the properties of paradise. We treat these properties as the meaning of paradise. However, there are few points that need to be addressed for benefiting from the. 政治大 Sandonicus approach is basically designed for English environment. To employ the 立. Sandonicus approach. First, this system will operate in Chinese environment and the. unofficial. Google. dictionary. API. to. do. so. (source:. ‧. ‧ 國. adopt. 學. Sandonicus approach, the words need to be translated from Chinese into English. We. http://code.google.com/p/google-api-translate-java/ ). Second, given a user is assumed to. sit. y. Nat. input a phrase to describe the goal, the phrase should be decomposed into processable. io. al. er. lexical units. This task is performed by using Chinese word segmentation API from. v. n. Sinica (source: http://ckipsvr.iis.sinica.edu.tw/). For example, if the user input (in. Ch. engchi. Chinese form:電影巨片) is “blockbuster movie”. i n instead U. of “paradise”, then this. phrase is first decomposed to two lexical units (i.e., blockbuster and movie) and then they are separately sent to Google dictionary to translate them from Chinese to English. After translation, the system then starts to send “as * as blockbuster movie” to Google Web query and gets a series of adjectives. Finally, it’s unavoidable to attain undesired results when we employ the sentence pattern “as * as vehicle” in Google. To settle this issue, we develop a two-step process to ensure we can get the quality results. The first step is to establish an exception word list. Any word in this list will be filtered out from the search results. For example, when we send a query “as * as 38.