最大化顧客參與行為於推薦平台：以品牌合作角度塑造達人知識 - 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. 政治大最大化顧客參與行為於推薦平台: 立以品牌合作角度塑造達人知識 ‧. ‧ 國. 學. n. er. io. sit. y. Nat. 於創新服務特色之心理偏好分析 Maximizing Customer Engagement Behavior through Recommender System: Framing Maven Knowledge awith v Perspective i l C Brand Alliance hengchi Un. 研究生:巫承安中華民國 104 年 7 月.

(2) 運用 Maximizing. Customer Engagement Behavior through Recommender System: Framing Maven Knowledge with Brand Alliance Perspective 於創新服務特色之心理偏好分析. by. Cheng-An Wu. A Dissertation Submitted in Total Fulfillment of. The Requirement for the Degree of. 立. 政治大 Master of Science. ‧ 國. 學. In. Management information Systems. ‧ sit. y. Nat. io. n. al. er. Supervisor: Soe-Tysr Yuan, Professor, MIS, NCCU. Ch. engchi. i Un. v. Department of Management Information Systems. NATIONAL CHENGCHI UNIVERSITY. July 2015. © Cheng-An Wu 2015.

(3) 致謝在研究所求學這兩年，對我實在受益良多，首先我要誠摯的感謝我的指導老師 - 苑守慈老師。帶領剛進研究所對論文懵懂無知的我，細心教導，用嚴謹的標準，卻又不厭其煩的一步一步地讓我深入了解服務科學的博大精深。老師的春風化雨，讓我得以完成這篇論文和完成研究所的學業，不僅如此，老師也時常幫助我們從生活上問題到未來的目標，非常感謝老師在這兩年來的教導。論文口試的時期，感謝宋同正老師和林宛瑩老師，不吝給予許多寶貴建議，使本論文能更加嚴謹和完整。也要感謝周思妤老師和呂伊婷老師，在我懵懵懂懂的碩一時，和我們一起討論論文方向和架構，良師益友的妳們，給予我十分寶貴的想法，讓我收穫豐碩，得以完成本論文。. 立. 政治大. ‧ 國. 學. 此外也感謝幫助我論文實驗的朋友和創業家們:Daisy、Vocals、茗熙坊、灣德文創、RIBOSOME，感謝你們的建議和參與，也感謝許多來參與實驗的受測者們，很抱歉我無法全部打上，你們的幫助和建議，才能使本論文完成。. ‧. sit. y. Nat. 研究所兩年也要感謝一路陪伴的實驗室朋友: 維正與志騰，一路上的互相鼓勵、討論、學習、幫助和互相捉弄使我的研究生活多了幾分色彩，也謝謝實驗室的學長姐們: 威辰、席筠、韻平給我們許多生活和學業上的建議和經驗. n. al. er. io. 分享，亦感謝實驗室的學弟妹們: 承豫、祖韻、曉晨、庭毅，除了幫助我們論文實驗外，實驗室也因為你們的淘氣增添了許多活潑快樂的回憶。還有謝謝和我一塊相處兩年的 102 級碩班朋友們，有你們的生活互動，和一些奇怪聲響，讓我的碩班生活多采多姿，你們的陪伴和回憶，日後必定會十分想念。. Ch. engchi. i Un. v. 最後，也是最重要的，誠摯感謝我的家人，謝謝你們在我學習的道路上，無論如何都給予最大的支持和鼓勵還有最溫馨的陪伴，因為有你們才能讓我一步一步地成長學習。希望日後的我可以不用再讓你們操心和擔心!. 巫承安 2015/7/28.

(4) 中文摘要在這個充滿繁多新媒體時代，使用者面臨到眾多資料和快速變動的環境，使用者在媒體的使用行為和選擇上更加依賴各種推薦平台的建議。除此之外，隨著社群媒體的興起，許多的推薦平台整合了社群的人們關係來提供更準確的建議和選擇。雖然推薦系統在影響使用者的使用行為有顯著的效果，然而企業和品牌卻鮮少去關注或了解如何增加顧客參與行為在整合社群媒體的推薦平台上。顧客參與行為並不只有傳統的交易行為，而是包含了所有直接和間接影響企業品牌的行為，像是使用者回饋、口碑傳播等。而且，現今尚未有清楚明確的定義哪些關鍵因素，會影響顧客參與行為在社群化推薦推薦系統，來藉此獲得顧客關注，形成正向生態系統。本研究中，我們根據達人在社群化推薦平台中具有重要的影響力的觀點，以促進重塑達人知識來改變原有達人的行為和態度，藉此影響所有一般使用者在社群化推薦平台的顧客參與行為。我們提出新的架構和系統來幫助中小型商家在推薦平台上影響更多的推薦達人，獲得更多的顧客參與。我們建立. 立. 政治大. ‧ 國. 學. ‧. 商家參與後台來幫助中小型商家可以洞悉達人的行為，我們也建立了重新塑造資訊的系統，提供達人所需要的訊息文章，藉此來改變達人的知識和行為。此研究發現，達人的行為會受到娛樂型、知識型和激勵型的文章訊息影響行為，一般使用者也會受到達人行為影響。此外我們藉由品牌合作角度來幫助得到更多的顧客參與行為，我們發現中小型商家可以在社群化推薦平台獲得. n. Ch. engchi. er. io. al. sit. y. Nat. 顧客參與且建立一個正向機制循環。. i Un. v. 關鍵詞：顧客參與行為、社群化推薦平台、品牌合作、重塑知識、達人.

(5) Abstract With the highly dynamic trend of service economy, the firms are increasingly to co-create value with brand alliance to advance their competition advantage. On the other hand, with the massive information on the new media, the referrals provided by recommender systems in combination with social media have significantly impact on customer behavior. In light of these trends, the markers and firms should aim to increase the customer engagement behavior (CEB) which goes beyond the traditional transactions including purchase and non-purchase behavior on social recommenders. In this research, we focus on the role of mavens who are powerful. 政治大. influencers on the social recommender. We propose a new conceptual framework for facilitating to impact the maven’s knowledge and behavior and increase the CEB on the social recommender for Small/Middle Enterprise (SME). We establish the SME support engagement site for increasing the CEB on social recommender and framing knowledge context to influence maven for achieving. 立. ‧ 國. 學. ‧. the insight of the maven’s behavior. As the result of research, we discover that maven engagement behavior would be influenced by the entertainment, information and incentive types in context from the brand alliance perspective and the non-maven are willing to be affected by maven behavior. Moreover, with. sit. y. Nat. n. al. er. io. this discovery, the SME can increase the customer engagement behavior on the social recommender. Ch. engchi. i Un. v. Keywords: Customer engagement behavior, Social recommender, Maven knowledge, Value co-creation, Multi-stakeholder..

(6) TABLE OF CONCTENT. CHAPTER 1 INTRODUCTION ............................................................................ 1 1.1 BACKGROUND AND MOTIVATION ............................................................................................... 1 1.2 RESEARCH PROBLEM ................................................................................................................. 3 1.3 RESEARCH METHOD .................................................................................................................. 5 1.4 PURPOSE AND CONTRIBUTION .................................................................................................... 6 1.5 CONTENT ORGANIZATION .......................................................................................................... 7. CHAPTER 2 LITERATURE REVIEW ................................................................. 8 2.1 CUSTOMER ENGAGEMENT BEHAVIOR ......................................................................................... 8 2.1.1 Customer Engagement Behavior matrix ........................................................................... 10. 政治大. 2.2 RECOMMENDATION TYPE ......................................................................................................... 12 2.3 FRAMING THE INFORMATION FOR DIFFUSION ............................................................................. 16. 立. 2.3.1 Framing theory ............................................................................................................... 16. ‧ 國. 學. 2.3.2 Summarization ................................................................................................................ 17. CHAPTER 3 IENGAGEMENT PROJECT ..........................................................19. ‧. 3.1 THE CONCEPTUAL FRAMEWORK OF IENGAGEMENT .................................................................. 20 3.1.1 Situation – Organization and Eco-stakeholders................................................................ 20. Nat. sit. y. 3.1.2 Organism – E-empowerment ........................................................................................... 21 3.1.3 Behavior – Customer Engagement Behavior .................................................................... 22. io. al. er. 3.1.4 Consequence - Value conversion...................................................................................... 23. n. iv n C 3.3 THE SYSTEM SCENARIO ........................................................................................................... 26 hengchi U 3.2 THE SYSTEM ARCHITECTURE OF IENGAGEMENT ....................................................................... 24. CHAPTER 4 MAVEN INFLUENCE FOR CUSTOMER ENGAGEMENT ON SOCIAL RECOMMENDER .................................................................................29 4.1 THE CONCEPTUAL FRAMEWORK .............................................................................................. 29 4.2 SYSTEM ARCHITECTURE .......................................................................................................... 33 4.3 BRAND ALLIANCE MODULE ..................................................................................................... 35 4.4 FRAMING MODULE .................................................................................................................. 37 4.5 MEASURE MAVEN ENGAGEMENT MODULE ............................................................................... 41 4.6 MEASURE CUSTOMER ENGAGEMENT MODULE .......................................................................... 42. CHAPTER 5 APPICATION SCENARIOS ...........................................................44 5.1 THE CONCEPT BEHIND THE SERVICE .......................................................................................... 44 5.1 SERVICE JOURNEY AND SCENARIOS........................................................................................... 44 6.1 PROPOSITIONS ......................................................................................................................... 52.

(7) 6.2 ASSUMPTIONS ......................................................................................................................... 53 6.3 EXPERIMENT DESIGN DETAILS ................................................................................................. 55 6.3.1 BACKSTAGE SUPPORT SERVICE IN EXPERIMENT 1'S DESIGN AND OBJECTIVE ............................ 55 6.3.2 FRAMING KNOWLEDGE SERVICE IN EXPERIMENT 2 DESIGN AND OBJECTIVE ............................ 60 6.4 EXPERIMENT RESULT AND DETAILS .......................................................................................... 65 6.4.1 THE RESULT OF EXPERIMENT 1 .............................................................................................. 65 6.4.2 THE RESULT OF EXPERIMENT 2 .............................................................................................. 68 6.5 THE INTERVIEW WITH THE MAVEN............................................................................................ 97 6.6 OTHER FINDING ..................................................................................................................... 101 6.7 DISCUSSION AND FINDING ...................................................................................................... 105. CHAPTER 7 CONCUSION ................................................................................. 109 7.1 CONTRIBUTION ..................................................................................................................... 109. 政治大. 7.2 MANAGERIAL IMPLICATIONS.................................................................................................. 110 7.3 LIMITATIONS AND FUTURE WORKS ......................................................................................... 113. 立. REFERENCE ....................................................................................................... 114. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i Un. v.

(8) LIST OF TABLE. TABLE 2.1 CUSTOMER ENGAGEMENT MAP MATRIC (FORRESTER 2008) ... 11 TABLE 2.2 THE OVERVIEW OF THE RECOMMENDER TECHNOLOGIES ......15 TABLE 4.1 CUSTOMER ENGAGEMENT DIMENSIONS. ....................................32 TABLE 4.2 BUILDING THE TOPIC PROCESS .....................................................37 TABLE 4.2 SETTING THE CONTEXT ...................................................................38 TABLE 4.3 MAVEN’S CUSTOMER ENGAGEMENT DIMENSIONS. ..................42 TABLE 4.4 CUSTOMER ENGAGEMENT DIMENSIONS. ....................................43 TABLE 6.1 SUBJECTS FOR BACKSTAGE SUPPORT SERVICE IN EXPERIMENT ........................................................................................................56. 政治大 TABLE 6.2 THREE PAGES立 FOR FRAMING KNOWLEDGE CONTEXT .............64. ‧ 國. 學. TABLE 6.4.1 THREE PERSPECTIVES ...................................................................65 TABLE 6.4.2 OPINION IN THREE PERSPECTIVES .............................................66. ‧. TABLE 6.4.3 QUESTIONNAIRE DESIGN .............................................................69. y. Nat. TABLE 6.4.4 DESCRIPTIVE STATISTIC ...............................................................72. io. sit. TABLE 6.4.5 HYPOTHESIS OF ENTERTAINMENT TYPE ..................................73. n. al. er. TABLE 6.4.6 ONE SAMPLE T-TEST ......................................................................73. Ch. i Un. v. TABLE 6.4.7 RESULT OF T-TEST ..........................................................................73. engchi. TABLE 6.4.8 SUBJECTS MENTIONS ....................................................................74 TABLE 6.4.9 DESCRIPTIVE STATISTIC ...............................................................75 TABLE 6.4.10 ONE-SAMPLE T TEST....................................................................76 TABLE 6.4.11 HYPOTHESIS OF INFORMATION TYPE OF COMMENTS .........77 TABLE 6.4.12 HYPOTHESIS OF INFORMATION TYPE OF RECOMMEND ......77 TABLE 6.4.13 DESCRIPTIVE STATISTIC .............................................................78 TABLE 6.4.14 ONE SAMPLE T TEST ....................................................................78 TABLE 6.4.15 DESCRIPTIVE STATISTIC .............................................................78 TABLE 6.4.16 DESCRIPTIVE STATISTIC .............................................................79 TABLE 6.4.17 HYPOTHESIS OF INCENTIVE TYPE ............................................80.

(9) TABLE 6.4.18 ONE SAMPLE T TEST ....................................................................80 TABLE 6.4.19 ONE SAMPLE T TEST ....................................................................80 TABLE 6.4.20 DESCRIPTIVE STATISTIC .............................................................81 TABLE 6.4.21 HYPOTHESIS OF COMBINE TYPE...............................................82 TABLE 6.4.22 ONE SAMPLE T TEST ....................................................................83 TABLE 6.4.23 THREE PAGES COMPARE .............................................................83 TABLE 6.4.24 DESCRIPTIVE STATISTIC .............................................................87 TABLE 6.4.25 HYPOTHESIS OF ALLIANCES BASED CAMPAIGN .................87 TABLE 6.4.26 ONE SAMPLE T TEST ....................................................................88 TABLE 6.4.27 PEARSON CORRELATION COEFFICIENT FORMULA ...............89. 政治大. TABLE 6.4.28 7 PEARSON CORRELATION COEFFICIENT RESULT ................90. 立. TABLE 6.4.29 DESCRIPTIVE STATISTIC .............................................................91. ‧ 國. 學. TABLE 6.4.30 HYPOTHESIS OF PROPOSITION 2 ...............................................92 TABLE 6.4.31 ONE SAMPLE T TEST ....................................................................93. ‧. TABLE 6.4.32 DESCRIPTIVE STATISTIC .............................................................94. io. sit. y. Nat. TABLE 6.4.33 HYPOTHESIS OF DIFFERENCE OF MAVEN AND NON-MAVEN..........................................................................................................94. n. al. er. TABLE 6.4.34 INDEPENDENT SAMPLE TEST ....................................................95. Ch. i Un. v. TABLE 7.1 ASSIST THE SME WITH OUR SUPPORT SYSTEM ........................ 112. engchi.

(10) LIST OF FIGURE FIGURE 2.1 DOORN’S CUSTOMER ENGAGEMENT CONCEPTUAL MODEL .. 9 FIGURE 2.2 A PROCESS OF MODEL ON FRAMING THEORY ..........................16 FIGURE 3.1 IENGAGEMENT CONCEPTUAL FRAMEWORK ............................20 THROUGH NEW MEDIA CHANNELS IN SERVICE ECOSYSTEMS ..................20 FIGURE 3.2 IENGAGEMENT SYSTEM ARCHITECTURE ..................................24 FIGURE 4.2. SYSTEM ARCHITECTURE. .............................................................34 FIGURE 4.3 BRAND ALLIANCE MODULES .......................................................36 FIGURE 4.4 SCORING THE SENTENCE PROCESS .............................................40. 政治大 FIGURE 5.1 SERVICE JOURNEY 立 ..........................................................................45 FIGURE 4.7 MAVEN BEHAVIOR DETECT...........................................................41. ‧ 國. 學. FIGURE 5.2 REGISTER PAGE ...............................................................................46. ‧. FIGURE 5.3 INTRODUCTION PAGE FIGURE 5.4 POSITIVE CYCLE ....................................................................................................................46. sit. y. Nat. FIGURE 5.5 OBSERVATION COMPONENT FIGURE 5.6 EXTERNAL EVENTS ..................................................................................................................47. al. er. io. FIGURE 5.7 LIST OF BRAND ALLIANCE STORE ...............................................48. v. n. FIGURE 5.8 CREATE PAGE FIGURE 5.9 CAMPAIGN THERMOMETER COMPONENT .........................................................................................................48. Ch. engchi. i Un. FIGURE 5.10 FRAMING MODULE .......................................................................49 FIGURE 5.11 FRAMING CONTEXT ......................................................................49 FIGURE 5.12 FIRM PAGE. FIGURE 5.13 INTERACT COMPONENTS ....50. FIGURE 5.14 FEEDBACKS ON CAMPAIGN THERMOMETER COMPONENT .50 FIGURE 6.1 ENTIRE EXPERIMENTS IN THIS RESEARCH................................55 FIGURE 6.2 ARCHITECTURE OF SUMMARIZATION SYSTEM ........................58 FIGURE 6.3 THE EXAMPLES FOR ALLIANCE BASED CAMPAIGN ................59 FIGURE 6.4 THE PROCEDURE OF EXPERIMENT 2 ...........................................61 FIGURE 6.5 INTERESTING DETECTION.............................................................62 FIGURE 6.4.1 SUBJECT IN EXPERIMENT2 .........................................................69.

(11) FIGURE 6.4.2 GENDER IN MAVEN FIGURE 6.4.3 INTERESTING IN MAVEN ...................................................................................................................69 FIGURE 6.4.3 ENTERTAINMENT TYPE ENGAGEMENT LEVELS ....................72 FIGURE 6.4.4 INFORMATION TYPE ENGAGEMENT LEVELS .........................75 FIGURE 6.4.5 IMAGE ABOUT WHOLE SERVICE FOR ENGAGEMENT .........79 FIGURE 6.4.6 COMBINE TYPE ENGAGEMENT LEVELS ..................................82 FIGURE 6.4.6 SCORE FOR WILLING ENGAGEMENT IN FIRM ........................84 FIGURE 6.4.7 INCENTIVE TYPE OF ENGAGEMENT .........................................85 FIGURE 6.4.8 ACTIONS OF MAVEN ....................................................................85 FIGURE 6.4.9 PAGE PREFERENCE ......................................................................86. 政治大 FIGURE 6.4.11 INFLUENCE 立LEVEL OF ENGAGEMENT BEHAVIOR BY. FIGURE 6.4.10 PAGE PREFERENCE OF GENDER ..............................................86. MAVEN ...................................................................................................................92. ‧ 國. 學. FIGURE 6.4.12 INFLUENCE LEVEL OF ENGAGEMENT BEHAVIOR BY MAVEN ...................................................................................................................93. ‧. FIGURE 6.4.13 POSITIVE FUNCTION DISTRIBUTION ......................................96. Nat. sit. y. FIGURE 6.4.14 COMMENT FUNCTION DISTRIBUTION....................................96. al. er. io. FIGURE 6.4.15 WILLING OF ACCEPT THE SERVICE .........................................97. iv n C FIGURE 6.5.2 NUMBER OF THEhSUBJECTS 102 i U e n g c h..................................................... n. FIGURE 6.5.1 ACTIONS ON THE PAGE ............................................................. 102. FIGURE 6.5.3 ACTIONS ON THE PAGE OF MAVEN ......................................... 103 FIGURE 6.5.4 ACTIONS ON THE PAGE OF SUBJECTS .................................... 104 FIGURE 6.7.1 FRAMEWORK OF PROPOSTIONS.............................................. 107.

(12) CHAPTER 1 INTRODUCTION 1.1 Background and Motivation. Over the past years, as the variety of new media increased, the ways consumers acquaint and make decision on goods have transformed dramatically (Hennig-Thurau, 2010). User-generated content on new media platforms and product search engines has been influencing consumer behavior for goods comprehensively (Ghose 2012). However, with the information explosive. 政治大 the information filtering, for example user referral or curation information, 立. growth people face a problem of information overload. To solve this problem,. ‧ 國. 學. becomes more persuasive. Because of this trend, recommender systems become more widely used, for example Ipeen, Tripadvisor, etc. (Prasad, 2012).. ‧. Recommender systems not only collect user experience information, but. Nat. sit. y. also gather the optimal amount of knowledge information obtained from mavens. n. al. er. io. who are the expert in some categories. On the other hand, as the more users. i Un. v. using the Internet and the social collaboration increasing, the more this trend of. Ch. engchi. using trust-based social recommender systems is perceived. Trust-based social recommender systems provide two notions of the trust: trust about the other users of the recommender and about to the recommendation site (Prasad, 2012). Mavens, who have abundant information about products, peripheral products and other aspect of market in the particular fields, are highly trusted and likely diffusing the information to other customers and deeply involve in particular market fields, even in categories of product that are not used by them (J. Boster, 2010); therefore, mavens play an important characteristic in social recommender systems. Including the maven’s knowledge and the curation information that 1.

(13) recommender systems are argued to influence on consumer’s cognitive and behavior (Thorsten, 2010). Social Recommender system is one of the technologies having drastically impact on customer searching and decision making on goods (Haubl and Trifts, 2000). With the trust of user knowledge and advice, firms can use such technology for providing highly individualized services and products based on numerous data online, such as searching engine and social media (Hennig-Thurau, 2010). On the other hand, customer engagement concept applied comprehensively in market place. (Van Doorn, 2010) Customer. 治政 engagement defined as the customers’ behaviors大 go beyond transactions and 立 customers’ behavioral manifestations have a brand- or firm-focus. Customer ‧ 國. 學. engagement behaviors are resulting from motivational drivers (cf. MSI, 2010). motivational. drivers. include. word-of-mouth. ‧. Such. activity,. sit. y. Nat. customer-to-customer (C2C) interactions and blogging activity. Therefore, based. io. er. on the customer engagement concept, the social recommender systems are. al. n. supposedly to be an appropriate tool to engage customer behavior to acquire the. ni C hbrand. potential customer for firm and U engchi. v. Nevertheless, it has been a gap exists between social recommender system and customer engagement behavior. Actually, when marketers engage customer on new media; most of them focus on social media interaction, for example Facebook, Twitter. However, seldom of them explore the possibility of using the recommender system to engage customer behavior. Senecal and Nantel (2004) claimed that recommenders have capability of altering and building up the customers evoked set. Integration of consumer preferences, such as the following the maven’s topic or the individual’s history records, in recommendation and consumer acceptance in recommender systems can affect 2.

(14) consumer decision making. This type of customer engagement should be connected to the social recommendation system by which all the new ways in which customers can interact with ﬁrms, including purchase and non-purchase behavior (Libai, 2011). Further, there are variety characters interacting with each other in the social recommendation system. For instance, firms manifest the brand and product on the social recommender system; customers search for the advice for the goods; social recommender system owners organize the both firm and customer on the site, and also held some events and activities. In order to. 治政 integrate this complex relationship, we expect to大 establish a digital business 立 ecosystem to build a mutual co-creation relationship between firm, ‧ 國. 學. recommendation platform and the customers/mavens on the platform (Omar A.,. ‧. 2013).. y. Nat. n. al. er. io. sit. 1.2 Research Problem. Ch. i Un. v. Since social recommender systems are argued to be suitable for engaging. engchi. customer behavior but it is rare for firm notice on this new media, this study aims to integrate recommender system with the engagement elements, i.e. the customer involvement, interaction, intimacy, and influence (Forrester, 2008), in order to influence customer attitudes and decisions and affects maven by diffusing the ideas or knowledge about firm. Review of previous research to recommender systems or platforms nowadays, these new media was often used only for searching or filtering information (Prasad 2012), seldom of them have marketing strategy for acquiring and developing customers. We believe that not only the social media 3.

(15) has the chance to engage customer behavior to influence other customers by E-WOM, but also the social recommendation system has the opportunity to engage customer behavior to referral customers by the abundant of maven’s knowledge and customer experience. Besides, Doorn (2010) addressed that firms influence CEBs by developing and providing processes and platforms to support specific customer actions, such as browsing the webpage, sharing the topic and referring the product. Accordingly, these studies will integrate the searching and sharing process on the recommender site to engage the customer behavior. In the social recommender. 治政 system, customer making decision on the goods via大 the scores and users’ advices 立 provided by customers on the page; in other words, the most persuasive ‧ 國. 學. comments are the knowledge and experience sharing by the mavens.. ‧. In this research, we focus on the question:. sit. y. Nat. 1. How firms gain the customer engagement behaviors on the social. io. al. er. recommender site by mean of maximizing the likelihood of the designated. n. maven knowledge behavior on recommender site compared with the traditional way.. Ch. engchi. i Un. v. 2. In addition, this research also considers the issue about if the framing context with different content type (Entertainment, Information, Incentive) would impact the maven engagement behavior on social recommender? With these two the issues, we would like to build an ecosystem, which includes the focal firm, brand alliance and customer on the recommender site, in order to obtain the positive engagement cycle (that integrates customer empowerment, e.g., customer comments, customer sharing, with firm strategy, in order to increase more customer engagement and more customer behavior). 4.

(16) 1.3 Research Method. In our vision, we would like to build a mechanism that can take in account well-design ecosystem to assist firms to engage the customer with the recommender system/site. In our mechanism, we focus on improving the interactions with customer on recommender. According to previous research (cf. MSI 2010, Thorsten, 2010), the most important factor for increasing the CEBs on social recommender system is maven’s knowledge sharing. Therefore, our service system would. 政治大. gather the information which is highly related to a focal firm from different. 立. media, for example social media, news, blog, etc. After that, those information. ‧ 國. 學. will be sent by our system to the mavens who are potentially interested in to advance the maven’s attitude and acknowledge, in order increasing the. ‧. likelihood of sharing.. y. Nat. io. sit. On the other hand, our mechanism adopts the ecosystem perspective and. n. al. er. considers both focal firm and stakeholders. The mechanism would utilize the. Ch. i Un. v. framing theory (Dietram, 1999)to cluster external source events and then connect. engchi. to the stakeholder relationship, and summarize the relative information. Framing theory is based on the mass communication on the media effect, which causes how to spread the particular information on the mass media or individual with the process of Frame building, Frame Setting and Individual-level effect of framing (Dietram, 1999). With the arrangement of information and specific issue, we believe it is mapping to the specialist sphere directly. To engage the customer behavior from recommender system to focal firm, our mechanism would build an engagement site to integrate all engagement modules. The engagement site can centrally control the customer behavior 5.

(17) analysis for the information to make the decision for choosing the product. Furthermore, the engagement site can include not only the recommendation modules, but also other channel modules, for example email module, social media module, search engines module, so that we pool marketing channel information and both existent and potential stakeholders, to formulate a positive ecosystem.. 1.4 Purpose and Contribution. 政治大. In our research, the purpose is to come up with a mechanism to. 立. accomplish the customer engagement through recommender system. In addition,. ‧ 國. 學. this mechanism enables value co-creation within ecosystems, which include the focal firm, stakeholders, and other media. Our mechanism is semi-automatic to. ‧. gather information and sensor possible external event on the recommender site. y. Nat. io. sit. to advance the customer engagement behavior.. n. al. er. Customer engagement concept is already mentioned on marketing (Harvey. Ch. i Un. v. 2005). Since the new media growth significantly, firms are focusing customer. engchi. engagement with social media, and lots of research (Thorsten,2010, Roderick, 2011) has concentrated on it. We would start to dedicate on recommender system and discover the method and system to support it. Besides, our mechanism regards the ecosystem perspective that can improve the value co-creation with different actors. We aim to connect concept of customer engagement on the recommender system with ecosystem through the engagement site, and fulfill the co-creation with firm and each stakeholders. As the result of the firm would considerate customer engagement concept and. 6.

(18) co-creation with ecosystem when design the business model and marketing strategy.. 1.5 Content Organization. In Chapter 1 is to introduce of our research background, motivations and define the research problem and purposes. In Chapter 2, which is the literature review section, we briefly discuss and detect our research proposition with the theoretical which is support us define. 政治大. the to reach the specific objectives that are extended from Chapter 1 and build. 立. the foundation of research knowledge.. ‧ 國. 學. This research is the partial from iEngagement project. Thus, in Chapter 3, we are going to show the entire project and purpose. Meanwhile, identify the. ‧. position of this research in the project.. y. Nat. io. sit. In Chapter 4, we developed a conceptual framework from the previous. n. al. er. review and finding related on the Chapter 2 and tried to establish architecture. Ch. i Un. v. with information technology to develop a possible solution that can be realized.. engchi. In Chapter 5, we have shown a conclusion and future work.. 7.

(19) CHAPTER 2 LITERATURE REVIEW. In this section, we figure out how to facilitate customer engagement through recommender, we review some existing knowledge to properly position the research. First, we overview the customer engagement behavior concept, identified the key factor on our research. Second, we review the popular recommendation type used in recommender, and identify the suitable type of recommender and discover the suitable factor or person to influence the customer engagement on recommender of our research. Third, we overview. 政治大. knowledge information spreading formation and define the type of knowledge to. 立. diffusion through the used acceptance and satisfaction. Then, we briefly. ‧ 國. 學. introduce and discuss the popular technology used in summarization today, and defined the suitable technology.. ‧ y. Nat. n. er. io. al. sit. 2.1 Customer Engagement Behavior. i Un. v. Customer engagement behavior (CEB) first thinks in marketing to. Ch. engchi. consider of value- creating with customer behavior that generate the corporate performance of firm financial (Brodie, 2011). A customer engagement behavior (CEB) is defined as the customers’ behavioral manifestation with a firm or brand focus, and go beyond the purchase, resulting from motivational drivers (Doorn, 2010). Customer engages to the firm with customer-to-firm relationship experience, which may be impact on motivational drivers both in positive and negative (see the figure 2.1). In firm perspective, firm can co-creation with the customer to co-create their service experience. As the relationship on firm and customer, CEBs can relative with the service-dominant (S-D) logic (Vargo and 8.

(20) Lusch 2008a), which emphasize the empowerment of customer and co-creation on service value.. 立. 政治大. ‧. ‧ 國. 學 sit. y. Nat. Figure 2.1 Doorn’s customer engagement conceptual model. n. al. er. io. On the other hand, as the technology growing and the trend of used on. i Un. v. new media, the influence of customer relationship on new media enhance. Ch. engchi. significantly. People communicate with each other for exchange the information about product or service experience more easily (Hennig-Thurau, 2010). With the customer interaction with each other and firm on the new media, firm are forced to measurement the impact and understand the interaction on each media, for example, in automated recommendation systems provide collaborative filtering service that impact when customer search the specific categories (Goldberg et al. 1992), the social media provide the E-WOM to impact the customer attitude when they obtain the information (Abrantes, 2012).. 9.

(21) 2.1.1 Customer Engagement Behavior matrix. Measurement the Customer Engagement Behaviors (CEBs) are the importance of the identification of the relationship with the customer and withers the service improves or not. According to Doorn’s research (2011), CEBs can separate in five dimensions: Valance, Form/ modalities, Scope, Nature of impact, Customer goals, can use to manage for identify, evaluate and react. In Brodie research (2011), he has summarized the Engagement Dimensionality; most of them distinguish in the multidimensional: cognitive, emotional, and behavioral. 政治大. perspective of engagement. In analytical aspect, in order to measure the CEBs go. 立. beyond the transaction, versus the traditional purchase, Bijmolt (2010) provide a. ‧ 國. 學. three key stage of modal for engagement for Acquisition, Development, Retention in analytics for customer engagement life cycle. Kumar (2010). ‧. provide concept to measure the Customer Engagement value (CEV) through. y. Nat. io. sit. customer life value, customer referral value, customer influence value and. n. al. er. customer knowledge value and also including the three dimensions: Behavioral,. Ch. i Un. v. Attitudinal and Network in order to capturing the customer engagement value.. engchi. On the other hand, Forrester (2008) build a customer engagement metrics to measure the popular data on the new media with Involvement, Interaction, Intimacy and Influence.. 10.

(22) Table 2.1 Customer Engagement Map Matric (Forrester 2008) Dimension. Definition. Metrics. Involvement. The visit of a person at the. Frequency of Web site visits. various firm channels or touch point, including website, physical channel. Interaction. The action that a person does. Average page views per visit. at the key touch point, include. Web site logins per customer. 政治大 reading content, page views. 立. Average time spent per Web. The emotion of like affection. Monitoring of customer. or aversion that a person holds. complaints. ‧. for a brand.. Opinions expressed in customer service calls. sit. y. Nat. al. n. advocate on brand by his. Likelihood to recommend. er. The likelihood of a person will. io. Influence. site visit. 學. Intimacy. ‧ 國. the completing transaction,. i n C U individualhbehavior. e n g cInhouri. v. research, The behavior with influence can be just providing feedback on the recommender. As our research purpose is to measure the customer engagement behavior on the new media, we used the Forrest’s four dimensions matrices to measure the action on the new media and combine with the Bijmolt’s three stages to analytics the different stage that new media can influence. 11.

(23) 2.2 Recommendation type. With the advances in information technology and the used in the Internet have accelerated the diversities and various alternative items in different domain in online, like thousands of songs, restaurant and hotel etc (Burke 2002). When the collections are increasing and information overloading, individual is force to time consuming on choosing the item. Recommender systems are design for solving these problems with friends and experts/mavens who have the knowledge about the item/product. Recommender systems most applied on. 政治大. e-commerce service in several categories (Nenkova, 2012). In general point view,. 立. Recommender system are help for suggesting the suitable items that user are. ‧ 國. 學. interesting, meanwhile, benefit both the user and the item provider. Recommender systems have been an important application and start forcing of. ‧. considerable recent academic like customer behavior area (Nenkova, 2012).. y. Nat. io. sit. Recommender systems utilize several technological methods to archive. n. al. er. the recommend task. Most of the commonly technologies used in Recommender. Ch. i Un. v. systems today can generally categorized into content-based filtering,. engchi. collaborative filtering, demographic filtering, hybrid recommender systems, trust based social recommender systems, agent based recommender systems (Nenkova, 2012). We reviewed all of these technologies and compare each other, for find out the most suitable technologies for impact on customer engagement behavior. Content-based filtering method utilizes the description or content of items to filtering and recommend. It used to item-to-item relationship to bridge the user needs and performance. In order to enhance the direction on recommendation on item, content-based system processes information from 12.

(24) various resources and needs to extract the useful characteristics and elements about its content. Each recommender would define its own description and element to facilitate the relevance of items. The advantage of content-based filtering is no need of the user historical data. Without using the user rating, they are able to recommend new item and unpopular item with less user rating to customer. On the other hand, the disadvantage is difficulty to analyze the multimedia content and finding something unexpected without any categorize (Burke 2002). Collaborative filtering uses the collections of the rating on a list of items. 治政大 by the user as a rating for user suggestion. Opinions can be allowedly given 立 score or can be explore from the historical data of the user. Most of the ‧ 國. 學. Collaborative filtering divide into two categories: User-based and Item-based.. ‧. User-based cumulate the correlation with other user and collecting the scoring. sit. y. Nat. information from user. With this method, user-based collaborative filtering. io. er. predicts the scores of the unrated items according to the historical rating data; the. al. item-based collaborative filtering predicts the score of the item by averaging the. n. iv n C current user’s rating data of similar in the past, item with sparse data is less h eitems ngchi U. important,. With this matrix, the item-based cause fewer problems with cold start and attacks comparing with the user-based. Generally, collaborative filtering does not need a representation of items of feature, only based on the participant and involvement of user community. This brings the customer behavior inside the chosen, but still conflict in different individual with different thought (Nenkova, 2012). Demographic filtering is to build models by other past user with clustering the stereotypes and the characteristics. A typecast is a collection of the characteristics and knowledge which frequently use by users or user groups. The 13.

(25) purpose is to quickly setting the new customer into a relative typecast, that can increase the directly of the recommend. The weakness of the demographic filtering is that model recommended based on clustering some similar characteristic, for example interest, the system might provide recommendation which is too general. Hybrid recommender systems combine two or more technologies to gain better benefit with other advantage; this is another category of recommender systems to overcome the limitations of the other approach. Most common hybrid recommender is to combine content-based and collaborative filtering to enhance. 治政大and user rating perspective. the relative of the recommend on content similarities 立 Hybrid recommender system including various types: Weighted, switching, ‧ 國. 學. Mixed, Feature combination, Cascade, Feature augmentation and Meta-level. To. ‧. fulfill the dynamic e-commerce environment, hybrid recommender systems is. sit. y. Nat. suitable solution to adapt the environment needs (Göksedef, 2010).. io. er. Trust based social recommender systems integrate with the user. al. community. The trust factor established in the user will be aware of the nature of. n. iv n C the recommendations. Based on from information about user profiles h ederived ngchi U and relationships between users, trust based social recommender systems emerges some rules that can be played by explanations in RSs, for example, trust , satisfaction, persuasiveness. In the Trust based social recommender systems, system provide recommendation based on the human interaction emotion and behavior (Nenkova, 2012).. Overall, after reviewing the several technological, we discover that the trust based social recommender systems establish by the human behavior versus the other technological with functional-oriented of few interactions. For the purpose to influence on user behavior, we expect that our research can have the 14.

(26) stronger effect on the customer engagement behavior on the social recommender systems media. Table 2.2 the overview of the recommender technologies Technique. Content-based. Collaborative Filtering. Demographic. Trust based social. Filtering. Recommender Systems. Item, Description. (for referral). Rating data. Item. Typecast. User Knowledge,. from user. Information. User group. User Suggestion. 立 Feature. User Matrix. ‧ 國. Item Element,. Item. Features. 學. Knowledge. Matrix. ‧. In the previous section, we learned that the trust-based social. Nat. y. recommender system has the highly potential to impact the customer. sit. Representation. Transaction data 治政大. io. engagement on maven’s behavior since the trust-based social recommender is. n. al. er. Input factor. i Un. v. recommended by user knowledge. Maven are described as ‘‘individuals who. Ch. engchi. have information about many kinds of products, places to shop, and other facets of markets, and initiate discussions with consumers and respond to requests from consumers for market information” (Feick and Price, 1987) .However, the mavens are deeply involved in a wide range of categories and understand the information about the service or detail of product even they do not use (Boster, 2011). Maven's influence is based on more general market expertise, besides, mavens also like to discuss the deals they get and frequently volunteer advice to others regarding purchasing decisions. Typically mavens are recognized by others as such and are sought after by others for information (Boster, 2011). 15.

(27) 2.3 Framing the information for diffusion. In order to impact the maven cognitive, we use the framing theory to diffusion the knowledge and using the summarization and ConceptNet to establish the knowledge.. 2.3.1 Framing theory. Framing first referred as a scattered conceptualization on the 1993. 政治大 (1999) defined framing. (Entman) first come from the study of mass communication. After integrate. 立. several research, Scheufele. as a concept on how. ‧ 國. 學. individual, group, and society to communicate about reality. Framing most applied on the media effect research and social theory area. Scheufele developed. ‧. a process model and integrate numerous researches on framing of media effect:. Nat. sit. y. frame building, frame setting, individual-level process of framing, and a. n. al. er. io. feedback loop from audience to journalist (figure 2.2).. Ch. engchi. i Un. v. Figure 2.2 a process of model on framing theory 16.

(28) The first step is framing building which similar from agenda-building by observation of the important of key structure and factors on the media in the people’s mind. Second, the frame setting is providing more detail and specific salience issue can impact the audience most that relative with the frame can transform the audience minds. The third step is the individual-level effects of framing. In this step, people adapt to the new behavior. The last step is a feedback loop from personal to mass media effect. It will impact the suitable cognitive and behavior that helps people change (Scheufele, 1999). To establish on the knowledge, we sort out the relative information that have appear on the. 治政 recommender using the framing theory to diffusion大 the knowledge. We integrate 立 the summarization tools and ConceptNet to arrange the information. ‧ 國. 學 ‧. 2.3.2 Summarization. y. Nat. io. sit. In the summarization are, there are numerous approaches for identify. n. al. er. important content for automatic text summarization. In generally, it separate in. Ch. i Un. v. two parts: Topic representation and Context influence (Nenkova, 2012). In the. engchi. topic representation part, it is commonly applied in Frequency-driven Approaches which including Term Frequency*Inverse Document Frequency, Latent Semantic Analysis, Bayesian Topic Models. Term Frequency*Inverse Document Frequency (TF*IDF) is beneﬁcial to be able to compare the continuous weights of words and determine which ones are more related to the topic and also this method can integrate with numerous documents.. (1). 17.

(29) Latent semantic analysis (LSA) is a robust unsupervised technique. Usually use for deriving an implicit representation of text semantics based on observed co-occurrence of words. It has applied in single and multi-document (Gong and Liu, 2001). (2) Bayesian Topic models are the most complex approach for topic representation proposed for summarization which has been steadily gaining popularity. The topic model representations are quite attractive because they. 政治大 research document resource based 立. capture information that is lost in most of the other approaches (Nenkova, 2012). However, since our. on the short and. ‧ 國. 學. comment-oriented message on the recommender system with lots of noise words inside, we select a basic and quickly method which is detecting the firm’s. ‧. important attributes as the critical topic for example: images or industry.. sit. y. Nat. n. al. er. io. In the sentence selection, we use the Term Frequency*Inverse Document. i Un. v. Frequency which can summarize the articles with critical words (Nenkova,. Ch. engchi. 2012). This method can improve the multi-documents and support in Chinese language, however, as the purpose of our research, we gather the information based on the message and comments, we combine the method with the positive and negative words and image describe words, which considerate the comment into the summarizations to improve the use of the comments information.. 18.

(30) CHAPTER 3 iEngagement PROJECT. The purpose of the chapter is to introduce the overview of our entire research project – “iEngagement”. Meanwhile, identify the proposition of our research in this project. iEngagement project facilitate integration of engagement in new multi-media. The goal of iEngagement research project is helping the firm achieve the maximization of Customer Engagement Behavior (CEB), such as WOM, Referral, and Page visiting in the new media’s ecosystem. Accroding to the Hennig-Thurau research (2010), the new media and service can derive into. 政治大. numerous types. Our iEngagement team combines these new media into three. 立. new media: Search Engine, Recommender, and Social media, and cover the most. ‧ 國. 學. of the new media and service. iEngagement project develop an concept framework that facilitate the value creation based on the CEB on the new media. ‧. with service system. Also our project is focus on the recommender part. We. y. Nat. io. sit. establish the detail according to the conceptual framework of iEngagement. n. al. er. research project (Chou & Yuan, 2010).. Ch. i Un. v. The following section is going to introduce conceptual framework in. engchi. section 3.1. Then, the system architecture of the project is shown on the section 3.2. The scenario of our system module is illustrated on the final section 3.3.. 19.

(31) 3.1 The Conceptual Framework of iEngagement. 治政大Framework Figure 3.1 IEngagement Conceptual 立 Through New Media Channels in Service Ecosystems ‧ 國. 學. In the figure 3.1, we provide our IEngagement conceptual framework. We. ‧. identify CEB value creation in four processes: Situation, Organism, Behavior,. y. sit. io. er. sections. Nat. and Consequence (Chou & Yuan, 2010). Each process is describe as following. al. n. iv n C Organization Eco-stakeholders h eand ngchi U. 3.1.1 Situation –. With the trend of digital service ecosystem, there are faced to fast-paced and the turbulent (El Sawy & Francis, 2013). Enterprises should position themselves as a federation of capabilities that co-create with other firm as a “business ecosystem." (Cherbakov et al., 2005; El Sawy and Francis, 2013). “Business ecosystem" is described an extended system of mutually supportive organizations. He defined it as ‘‘communities of customers, suppliers, lead producers, and other stakeholders interacting with one another to produce goods and services’’ (Moore, 1998). 20.

(32) The all players are interact with that of the ecosystem this represent the individual must compete as a value-adder and that the number of competitors may be quite different in a value network to those in a value chain. This combination of cooperative and competitive processes has been termed “co-opetition’’.. 3.1.2 Organism – E-empowerment. The digital systems, technology and artifacts are key components they are. 治政 the main important player as the transformation 大 and a key component of the 立 ecosystem. Nowadays, the trend of the rise of new media channels such as ‧ 國. 學. Facebook, Google, and Twitter, which transform customers to take a more. ‧. interaction role as market players and reach (and be reached by). Accroding to. sit. y. Nat. the Hennig-Thurau et al. (2010), we define the new media as websites and other. io. er. digital communication and information channels in which active consumers. al. engage in behaviors that can be consumed by others both in real time and long. n. iv n C afterwards regardless of their h spatial location. However, after the rise of new engchi U. media, the research regarding consumers with extensive options for actively providing information on service has been received attention (Hennig-Thurau et al., 2010). Therefore, new media also empowered consumers to privide and distribute their own contributions (Hennig-Thurau et al., 2010). In this framework, we refer e-empowerment as to the empowerment perceived by customers on the new media toward the firm.. 21.

(33) 3.1.3 Behavior – Customer Engagement Behavior. According to the previous section, the significance of e-empowerment is perceived by customers toward the firm impact on CEBs in a broader network of customer, firm and other stakeholders (Hennig-Thurau et al., 2010). Customers have the chance to show their attitude and behavior to the firm and its eco-stakeholders, since they increasingly participate in the creation, production, and delivery of service (Zeithaml, Bitner, and Gremler 2009). Drawing from role theory (Kahn et al. 1964), Verleye et al. (2013) assumed that customers’. 治政大its stakeholders depends not willingness to show CEBs that benefit the firm and 立 only on customers’ affect toward the firm but also on their role readiness.Based ‧ 國. 學. on the Verleye et al. (2013), the customer role readiness is defined as the degree. ‧. to which customers feel prepared for encounters with the organization in terms. sit. y. Nat. of feeling confident and having the appropriate knowledge and skills. Previous. io. er. research merely demonstrated that customer role readiness affects their. al. n. compliance with the organizational rules and procedures in various industries. i n C2005). (Auh et al. 2007; Meuter et al. U hengchi. v. Verleye et al. (2013) proposed that customers who do not have the appropriate knowledge or skills for encounters with the organization might be less willing to feedback to service system, or give suggestions for service improvement. Thus, we identify that e-empowerment perceived by customer increases customer role readiness, resulting in higher levels of CEBs.. 22.

(34) 3.1.4 Consequence - Value conversion. The rising of using the digital media, the boundaries will be more difficult to define where the organization ends and the other parts of the ecosystem begin (El Sawy and Francis’, 2013). At the same time, the notion of presumption will provided as consumers of services and products engage in their production through processes that can impact on firm traditional process such as user-generated content. Thus in a digital service ecosystem, value is co-created, co-converted, and co-captured together with the different players in the. 治政大and community. The value ecosystem: customers, competitors, complementors, 立 of a customer’s contribution and feedback to initiating new product and service ‧ 國. 學. innovation ideas needs to be included as a component of CE value. (Kumar et al.,. ‧. 2010).. sit. y. Nat. As a result, the extent to which customers are willing to engage in. io. er. conversations (with other customers as well as the firm) can significantly influence a firm’s value, especially as it affects what customers are prepared to. al. n. iv n C tell others, and what insights they to provide firms regarding product h earenwilling gchi U development and enhancement.. In manufacture perspective connect with the co-creation concept, with such customer participation, manufacturers have the potential to enhance product innovation and to speed up the development process, both of which are key objectives of managers to lower costs and improve market acceptance of new offerings (Athaide, Meyers, and Wilemon 1996). Thus, in a digital service ecosystem, value is co-created, co-converted, and co-captured together with the different players in the ecosystem: customers, competitors, complementors, and community Thus, organization in keystone positions in the ecosystem may 23.

(35) choose to leave many activities of value creation to others in the ecosystem, while choosing to focus on creating value that is critical to the ecosystem’s prosperity. Therefore, we argue that CEBs toward the organization increase the extent of value cocreation by customers; continuously resulting higher levels service offerings of organization and eco-stakeholders.. 3.2 The System Architecture of iEngagement. Stakeholder. CEB Cocreation. Social media. Search Engine. CEB Collection Module. io. er. Nat. Recommender System Module Search Engine Module. Recommender System. ‧. ‧ 國. Social Media Module. 學. Identification Module. Brand Partnership Module. y. 立. Support Data 政治大. sit. User Module. al. n. iv n C h Feedback engchi U Figure 3.2 IEngagement System Architecture Our system architecture provides the service to facilitate the customer e-empowerment perceived by customer increases customer engagement behavior on firm perspective. We use stakeholder identification module to identify the potential co-creator in the intertwined of the ecosystem. Besides, we use the three modules to e-empower the customers on new media, moreover, through CEB data from CEB collection module and co-create by brand partnership module.. 24.

(36) (1) Stakeholder identification module: In the digital ecosystem, firms are easily to co-create with each other, at least they find out the suitable partner. The purpose of this module is to give a list of recommend stakeholder that are potential cooperate with each other using data from the support database. (2) Social Media Module: The main objective of social media module is to facilitate the firm to maximize the possibility of consumers to do the empowerment behavior, such as content like, share and feedback. Through spreading on the social relationship, behavior can influence the customer effect according to social exchange theory and impact the acceptance as. 治政 customer role readiness, in order to give rise to大 the B2C and C2C CEB. 立 (3) Search Engine Module: Search engine is one of the most popular ways for ‧ 國. 學. customer to search the not familiar information. It is important to let the. ‧. target customers get firm’s information when the target customers use the. sit. y. Nat. key words that is related to the firm. The main objective of this module is to. io. er. help the firm to increase the acquisition of search engine. Through adjusting. al. the keyword and URL to obtain the acquisition, the firm can then established. n. iv n C the customer’s awareness of the brand and the knowledge for enhance the he ngchi U customers encounters with the organization (customer role readiness). (4) Recommender system module: As the information overloading, the importance on recommender system for filtering data has increasing. The main objective of recommender system module is to assist the firm to maximize the possibility of consumers by finding the maven (expert in the domain) to curate the knowledge content to further develop the customer’s role readiness. Through the development of the customer’s role readiness, firm can have an arousal on CEB. 25.

(37) (5) CEB collection module: CEB collection module purpose is gathering the matrices of CEB to provide the feedback on censor present state and support to the three new media module so that the three new media modules can dynamic adapted service to the specific firms. (6) Brand Partnership module: Brand partnership module’s objective is that the firm to communicate with the multi-stakeholder. We use the most common way in business: email communication channel, to help the firm contact with other the brand partnership and their cooperators. 3.3 The System Scenario. 立. 政治大. Considering with the customer engagement on the new media,. ‧ 國. 學. iEngagement team established a customer engagement site for provide the entire. ‧. service on the Taipeing website (http://www.taipeing.net/). Taipeing website is. sit. y. Nat. suitable to be centroid of the customer engagement site, because it contains more. io. er. than 20 thousands of stores with the relationship with stores. Besides, it contains. al. n. the image of feature on each store with matching the color image in 14. i n categories such as romantic, C relaxing, nature, etc. U hengchi. v. In our scenario, R8 bookstore is our focal firm, as the service user, needed to improve the customer awareness on the new media. Through our iEngagement system service, at the beginning, R8 utilize the stakeholder identification module to find out the suitable partner to cooperate. After filtering some partner, R8 decide to hold a joint camping with Elite bookstore. R8 utilize the our search engine module to manage the web page with search engine, recommender system module to improve the customer contact ratio and feedback on the recommender, and social media module to increasing the user behavior on the Facebook. The detail of the description of scenario is shown on as the following: 26.

(38) . Optimization of keywords and links of joint campaign pages through search engine module: R8 and their stakeholder, Elite bookstore, can submit some of description of a joint campaign pages for using SEO service: In NCCU (Location), Young people’s exhibition (Participants), Open exhibition spaces (Place), unfettered (Feature), there are new service on books for 20% sales. Optimization SEO service will provide R8 with some keywords – Wenshan, Young, Freedom – for them to improve their joint campaign pages’ content keywords. Besides, Optimization SEO service will also suggest R8 with. 治政大We expect that the use of some partners’ website links, such as NCCU. 立 these two features allow the page rank upgrade. ‧ 國. 學. . Influence maven cognitive & behavior through recommender system. ‧. module: R8 can provide us their marketing strategy with some option on. sit. y. Nat. iEngagement, for example hold joint camping with Elite bookstore. After. io. er. ensuring business and their stakeholders, iEngagement will collect the. al. relative articles on internet with both stakeholders and firm. Our. n. iv n C recommender system module find out the maven interesting with the h ewould ngchi U. firm on recommender. Recommender system module then establishes articles from historical reviews and sharing, in order to influence mavens’ knowledge through these framing summarizations. We expect we can improve the CEB on the recommender. . Maximizing introvert leaders WOM likelihood through social media module: R8 can provide a link of their Facebook fan page, then social media module analysis their fans’ behavior data and find out the introvert leaders – user A, B, and C – on their fan page. Social media module gives three introvert leaders the articles about R8 and the joint marketing are also 27.

(39) mentioned about. Using some articles that introvert leaders are interested in to maximize WOM likelihood and engage fans. Sharing the data of each module for progress the service is established on our architecture. The firm (R8) can manage their new media marketing and improve the customer engagement behavior on the iEngagement site. We are expected that we provide a centroid service in order to enhance the customer engagement behavior toward the firm.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 28. i Un. v.

(40) CHAPTER 4 Maven Influence for Customer Engagement on Social Recommender 4.1 The Conceptual Framework Figure 4.1 depicts the underlying conceptual framework of our mechanism, and interprets the relationships of the five important concepts, firm content type, stakeholder content type, framing knowledge, maven engagement and customer engagement, we also participate the engagement site concept into our conceptual framework. From the firm and brand perspective, the purpose is to advance. 政治大 engagement behavior as the 立. customer engagement, so that, we output some consequence and measurement about the customer. result on the social. ‧. ‧ 國. as follows:. 學. recommendation system. The conceptual framework is subsequently explained. Firm Content type:. sit Maven Engagement:. n. al. er. io Information Introduction History reviews. y. Nat. Entertainment Inner Event External Event. Ch. engchi. i Un. v. Customer Engagement:. Involvement Browse rate Involvement Store Browse rate. Framing Knowledge: Incentive Image Topic building. Brand Alliance Content type:. (1). Interaction Maven‘s Collections Maven‘s like. (2) Interaction Reviews. Intimacy Maven‘s Comments. Context setting. Entertainment Inner Event. Influence Maven‘s Share posts Maven‘s Comments. Incentive Image. Figure 4.1 Conceptual Framework 29. Intimacy Comments.

(41) Firm content type: People obtain information from various sources. All the information would impact the thinking and the behavior. According to the research of Irena (2013), based on the Uses and Gratifications (U&G) theory (Katz 1959), individual user would be impacted by different important factors on social media and technology when contact with the brand communities; for example the entertainment factor was found to increase the customer motivations on brand contact (Park et al. 2009) and information content with high entertainment was the main factor on online engagement behavior of individual (Dholakia et al. 2004; Raacke and Bonds-Raacke 2008). On the recommender, the user consults with the. 治政大 behavior (e.g., rating, knowledge, which includes the reviews and user historical 立 sharing) on recommender. Besides, the external source of curator would impact the ‧ 國. 學. user thinking and the behavior, for example, user curation content which is the. sit. y. Nat. suggestion.. ‧. arrangement of specific categorized information has often influenced brand. io. er. We accordingly classify these user content types into three parts,. al. entertainment, information and incentive. Information content factor is presented by. n. iv n C the knowledge, user behavior or curation Entertainment content is h e nong recommender. chi U the factor that is related to the activities on recommender or external place, for. example, voting events on recommender or the news. Incentive content is the image of feeling which can be associated with the firm. For example the romantic or relaxing feeling can attach to some restaurant. That is, we collect both internal and external information, activity event and news, which are held in recommender or the external media. Brand Alliance content type: In the ecosystem perspective, all of focal firms have stakeholders, complementors, competitors, customers and Community. Stakeholders are meant to co-create the value (Omar, 2013). The information of 30.

(42) product or firm will also impact by the interactive action of stakeholders; that is, we consider the cooperation with stakeholders as brand alliance content type, which also contains entertainment of campaign and incentive. Framing knowledge: In order to scatter the information of firm on the recommender system, we apply the framing theory (Dietram, 1999) to expect the opinion audience has media effect on the recommender; that is, mavens would access our incoming information and change their attitude. For this object, we create an article which is including the summarize topic and context of information that spread to maven as a reviewer who is willing to accept.. 治政大 1999), we build a frame First, according to the framing theory (Dietram, 立 based on the agenda-setting research by finding the salience issue, that is, the main ‧ 國. 學. topic from the historical content, for example firm content type and stakeholder. ‧. content type or other mass media. Then, we select the issue attributes from firm. sit. y. Nat. content type and stakeholder content type to set the frame as the article context.. io. er. Arrow (1): When the article spread to the maven on the recommender,. al. most of situation that information recipient would ignore the information which is. n. iv n C not relative with them; on the contrary, would access the incoming information h e they ngchi U. that they are interested in. On the other hand, mavens are more aware and interested in new product or the information than that of the non-mavens (Feick and Price, 1987). Besides, maven likes to share and suggest to the other people what they have learned. In order to examine what relevant information would influence the willingness of the maven’s acceptance and diffusivity to other people, we classified firm content type and stakeholder content type based on the Uses and Gratifications (U&G) theory (Katz 1959). Maven Engagement: In this section, we expect that mavens have some involvement and behavior to the focal firm since they receive our article with 31.

(43) integrated information by framing knowledge concept. To confirm that the mavens are really affected and engaged on the focal firm, it is required to follow the specific mavens to identify what their behavior have changed. We use Forrester’s engagement framework (2008), Involvement, Interaction, Intimacy, Influence to measure the degree of the maven’s behavior change, the definition is shown as follow table4.1. Table 4.1 Customer Engagement dimensions. Engagement dimensions. Definition. Involvement Engagement. The visit of a person at the various firm channels or touch point,. 立. ‧ 國. 學. ‧. completing transaction, reading content, page views. In our research, we concentrate on user-item rating actions. The emotion of like affection or aversion that a person holds for a. n. al. er. io. sit. y. Nat. Influence Engagement. recommender site but also other media channel. The action that a person does at the key touch point, include the. Interaction Engagement. Intimacy Engagement. 治政 including website, physical大 channel. In our research, we focus on. v. brand. In our research, we analyze the comment semantic to. i n C U hengchi rating.. The likelihood of a person will advocate on brand by his individual behavior. In our research, The behavior with influence can be just providing feedback on the recommender.. Arrow (2): On the recommender, the opinion influencer is the expert who is highly involved in some categories, that is, the maven. When the maven’s behaviors alter, we expect that maven would increase the discussion of the issue and firm’s information. Also, as mentioned before, the other non-maven users on the recommender would be impacted by the maven’s behavior. In this study, we would 32.

(44) find the level of engagement of customer that would be impacted by the maven’s engagement. Customer Engagement: Customer often influences by maven’s behavior, such as referral or advices in social recommender systems. (J. Boster, 2010). We believe that through increasing the maven’s engagement behavior on the focal firm or stakeholders will directly attain to create and increase the customer engagement behavior on the focal firm.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i Un. v. 4.2 System Architecture. Our system main purpose is building a firm perspective service on the social-based recommender for increasing the customer engagement to firm (see the figure 2). We use two parts to fulfill our objective: multi-stakeholder module, and framing. Firm user can use interactive operation to detect the information on 33.

(45) recommender and receive some advice to develop some strategies to increasing the customer engagement.. Historical Content. Multi-stakeholder Module. Building Topic. 立. Detecting Behavior. Identify Engagement level. ‧. Nat. io. Detecting Behavior. Recommender Data. n. al. y. Measure Customer Engage module. Identify Engagement level. er. Maven Data. 政治大. Spreading Information. Searching Maven. Article Data. 學. ‧ 國. Maven Engage module. Setting Context. Brand Alliance Data. sit. Framing Module. Advising Brand Alliance. Detecting Image. Sensing Position. Ch. engchi. i Un. v. Figure 4.2. System Architecture. In the system, we collect the firm historical content, user behavior and categories as a firm content type from the recommender, to set our database. Also, we gather other information of firm from the mass media. Through the multi-stakeholder module and operation filtering from firm user, we can get some advice of the most relevant stakeholder of image or issue as a stakeholder content type. These recommend stakeholder have high potentials to cooperate with the focal firm. After gathering the two content types on recommender, 34.

(46) content framing module creates the knowledge articles through the topic building and context setting that have the prospect to attract the maven’s awareness. In order to achieve the goal, it would consider the previous module information to collect the historical context and set a relevant framing that match the firm’s strategy or issue. Besides, the maven module connects with outside recommender for obtaining the data of maven and the categories on the recommender. And map the framing knowledge with the maven’s interest. Finally we detect the customer behavior level with the customer engagement module to identify our progress. The following paragraph will detail describe. 治政大 and explicate the operation of these modules. 立 ‧ 國. 學. 4.3 Brand Alliance module. ‧. sit. y. Nat. The purposes in the Brand Alliance module are to find the similar and. io. er. potential cooperator for providing the suggestion of co-branding or. al. co-advertising events on the recommender. The similar brand alliance (same. n. iv n C image or in the same market) might two possible relationships: partner, h e ninclude gchi U which in small firms would cooperate with each other to scale and sharing information, for example the traditional photo stores gather together in a street to. get the more popularity which is called economies of agglomeration; competitor, which in different size firm would compete each other have the opportunity to complement the resources with each other; for example the famous convenient store Circle K with lots of goods cooperating with 76 gasoline with huge placeholder to co-branding a new service. Co-branding and Co-advertising are defined as the combining and retaining of two or more brands to create a single product, service and advertisement (Kohli, 2002). 35.

最大化顧客參與行為於推薦平台： 以品牌合作角度塑造達人知識 - 政大學術集成

最大化顧客參與行為於推薦平台：以品牌合作角度塑造達人知識 - 政大學術集成