以網路口碑之語意萃取為基礎的推薦系統

A Recommendation System of Semantic Understanding Based on Consumer Generated Media

李佳曄(Jia-Ye Li)

指導教授 ： 戚玉樑

中原大學/商學院/資訊管理學系/碩士(2008年)

https://doi.org/10.6840/cycu200900486

摘要

口碑是資訊傳遞者與接收者面對面所產生的資訊溝通行為，這種商業溝通行為是非商業意圖的，為談論有關某一品牌、產品或服務的訊息傳播過程，是影響顧客消費決策的重要因素。隨著網路的發展，口碑不只限於以傳統口耳相傳的方式傳播，而漸漸將顧客的意見、經驗與評論等藉由消費者自主媒體(CGM)，如討論區、留言板、部落格等空間傳播，形成新型態的網路口碑。過多的網路口碑資訊，對人類來說不易閱讀，更別說透過網路口碑制定消費決策。因此本研究嘗試整合網路口碑資訊，解析商家評價，透過各項指標針對商家產品、服務做評論，透過各項指標比較商家之間的差異，對各類型消費者提出建議，建立出以網路口碑之語意萃取為基礎的推薦系統。本研究透過文字探勘(Text Mining)技術解析網路口碑，找出影響領域的關鍵詞彙，並透過萃取出的關鍵詞彙將非結構化電子文件，透過分析轉換成機器所能解讀的資訊。以知識本體(Ontology)建立店家的知識架構，透過語意網路規則語言(SWRL)將網路口碑納入知識架構之中，並經由文件量化處理，推論出各類型優質店家，並以摘要的方式呈現店家資訊，給予消費者查詢。

關鍵字

網路口碑 ； 消費者自主媒體 ； 文字探勘 ； 知識本體 ； 語意網路規則語言

並列摘要

Word-of-mouth is an information communication behavior resulted from transmitter and receiver face to face. This kind of commercial communication behavior is non-commercial intention. It discusses about some brand, product or the communication processes of information, and important factor that influences customer’s decision to consume. By the development of Internet, word-of-month not only limited communication by traditional word of mouth but also adopted Consumer Generated Media (CGM), liked customer’s opinion, experience and comment etc. Communication liked discussion, message board, blog etc, became new type of online word-of-mouth. The overloading online word-of-mouth information was not easy to read for people, needless to say that made consuming decision by online word-of-mouth. Thus, this study tried to integrate online word-of-month, analysis the evaluation of stores, criticism products and services by indicators. Comparing with the differences among stores and making recommendations to every type of customers by indicators, established a recommendation system of semantic understanding based on consumer generated media. This study analysis online word-of-mouth by text mining, found out key words that influence domains and made non-structure electronic texts transferred to machine readable information by extraction key words. Using ontology to establish the knowledge structure of stores and making online word-of-mouth subsume to the knowledge structure by Semantic Web Rule Language (SWRL). During the text quantification process, inferred every type of good stores and presented store information for customer checking as abstract type.

並列關鍵字

Text Mining ； Ontology ； Semantic Web Rule Language(SWRL) ； Online Word-of-Mouth ； Consumer Generated Media(CGM)

參考文獻

[9] Chen, K. J., and Bai, M. H. “Unknown Word Detection for Chinese by a Corpus-based Learning Method,” Computational Linguistics 3(1) 1998, pp.27-44.

[1] Schwartz, B.，劉世南譯，「The Paradox of Choice: Why More is Less (選擇的弔詭)」，台北: 天下雜誌出版，2004。

[2] Arndt, J. “Role of Product-Related Conversations in the Diffusion of a New Product,” Journal of Marketing Research 4(3) 1967, pp.291-295.

[3] Bickart, B., and Schinkler, R. M. “Internet Forums as Influential Sources of Consumer Information,” Journal of Interactive Marketing 15(3) 2001, pp.31-40.

[7] Brown, J. J., and Reingen, P. H. “Social Ties and Word-of-Mouth Referral Behavior,” Journal of Consumer Research 14(3) 1987, pp.350-362.