以語意相似度為基礎之線上購物資訊查詢方法

An Information Retrieval Method based on Semantic Similarity Degree for Online Shopping

在網際網路進行資訊查詢是電子商務活動一項重要的議題。至目前為止，大部分的資訊方法均未能有效處理語意資訊查詢之問題。有鑑於此，本研究提出一個以物件語意相似度為基礎之語意查詢法來解決上述問題，本研究係以兩個香水產品之間的精油、溶劑、酒精與水成份之相似度來計算香水產品之間的物件語意相似度，並以共識演算法發展一個出計算權重的方法。本研究為了證明提出此一方法在實務方面應用之可用性與實用性，發展出一個語意相似度的資訊查詢系統，並且採用關鍵字查詢法與語意查詢法進行香水產品之購物網站之資料集進行實驗，根據實驗結果顯示，本研究提出的語意查詢方法在類似產品查詢方面比關鍵字查詢法擁有較為優異表現。

Information retrieval (IR) on the internet is a key issue in electronic commerce (EC). However, most of IR methods can not deal with the semantic issues on IR. The purpose of this paper is to propose a semantic information retrieval method based on semantic similarity degree (SSD) to resolve the issues mentioned above. The computational method of SSD is derived from the concept of object class similarity concerning with the similarity of essential oils, solvent, alcohol and water between two perfume products. An algorithm based on the consensus algorithms is developed to compute the weight for the semantic similarity. To demonstrate the applicability and practicability of the proposed method, a semantic IR system is developed in this paper. An experiment using the proposed method and the keyword-based information retrieval method on the data set of a perfume online shopping website is conducted. Experimental results show that the proposed semantic IR demonstrates substantial enhancement of information retrieval quality in choosing similar product for online shopping.