使用模糊本體開發餐廳口碑的推薦系統

Using Fuzzy Ontology to Develop Restaurant Word Of Mouth Recommender System

時立哲(Li-Che Shih)

指導教授 ： 戚玉樑

中原大學/商學院/資訊管理研究所/碩士(2013年)

https://doi.org/10.6840/cycu201300938

摘要

中文摘要
知識本體提供知識庫一個完整的分類架構，能將生活中的明確知識模型化存入知識庫裡。但現實生活中存在著許多不明確知識，如感受、認知、主觀意識等；受限於系統在不明確資訊的表達，因此建置本體時須先透過人為判斷，將模糊知識轉換為明確資訊；但經由人為判斷反而與事實知識產生偏差，衍生後續的推薦偏差。另一方面，由於傳統以知識庫為基礎的推薦系統使用明確推論進行推薦，即使多數條件極為符合，只要有某一項目略低於推薦標準便遭系統剃除。
因此本研究嘗試發展一個以不失真為目標，處理不明確知識的「知識系統」，藉由知識觀點、知識關聯、模糊邏輯等學理，應用來解決實務問題之專家系統。本研究將明確的知識本體中加入模糊元素，成為模糊知識本體，使得明確推論中帶有模糊區間，推薦過程更貼近人類推薦過程，給予合適程度的彈性推薦和排序。本研究以餐廳推薦為研究議題，設計以模糊本體為基礎之餐飲推薦系統。

關鍵字

專家系統 ； 模糊理論 ； 知識本體 ； 推薦系統 ； 電子口碑

並列摘要

Abstract
Ontology provides an analysis structure to knowledge field, and will use specific methodology, to keep knowledge into a repository. However, lot of uncertainties fulfill in daily lives, such as emotions, recognition, subjective consciousness, and etc. Due to the uncertain expression of information in limited systems, building Ontology must be through human judgment. It will alter vague knowledge into accurate pieces of information; through human judgment, there are different between the scientific facts and the predispositional opinions of people. Because of these differences, there will be a turning deviation in future of ontology. On the other hand, a recommendation system based on traditional knowledge units will make utilize of the accurate conclusions to proceed recommend. Even though the most conditions match, the recommend standard will be left behind because of the inability to meet the recommended requirements.
Thus, this study aims to develop a system that utilized new technology to process uncertain information. Scientific viewpoints, connection, and the blurred theories have been utilized to practical problems for professional systems. This research has added the element of blurriness to the definite aspects of ontology. Results of our research show an unclear view on ontology, causing gaps of ambiguity to fall through the inferences of knowledge. The recommendation from the program is much closer than the ones from human, which offers an ideal result along with proper order. Our main research circles around the rightful direction for restaurants and anything among their field of work; the design of this program utilizes Fuzzy Ontology as a base for a capable approval to this field.

並列關鍵字

Fuzzy theory ； electronic Word-of-mouth(eWOM) ； Recommendation system ； Expert systems ； Ontology

參考文獻

陳榮靜、葉峻儒、李宜晏（2009）。以模糊正規化概念分析為基礎的本體論合併。資訊科技國際期刊，3(1)，35-55。

陳德美（2009）。發展語意感知模型於網路口碑推薦。（博士論文）。取自臺灣博碩士論文系統。（系統編號 097CYCU5396046）

蘇靖淑、洪久賢（2011）。台灣餐廳評鑑制度執行與發展之研究。運動休閒餐旅研究，6(1)，45-65。

戚玉樑（2006）。知識擷取與知識表達協同程序於建構本體的概念架構。資訊管理學報，13(2)，193-215。

許中川、陳景揆（2001）。探勘中文新聞文件。資訊管理學報，7(2)，103-122。