| Summary: | 碩士 === 元智大學 === 資訊管理學系 === 106 === With the prevalence of the Internet and mobile devices, physical channels have been replaced by e-commerce, whose product type, price competition, payment methods, and delivery time are prerequisite to consumers’ purchasing choices. The appearance of many e-commerce price comparison websites helps consumers to screen and commends e-commerce stores and product types. Every price comparison websites differ in mechanism for screening e-commerce products, thus influencing the speed and accuracy of product searching.
This study improves the information retrieval algorithm BM25 (Best Match) by adding in product name weighing. Search keywords were segmented and imported to Crawler to retrieve e-commerce product information and then to implement weighing analysis based on product names. After relevant analysis, the search result enabled users to browse products of the same type from multiple e-commerce stores. Since e-commerce product types have been diverse and have kept being updated, the price comparison system design was based on real-time retrieval and processing of product information. With JSP applied to Tomcat and Servlet for efficient allocation of product name weighing and of product price ranking. The experiment focused on the speed and accuracy of product search among famous price comparison websites, with four types of 3C products adopted as comparison keywords. With four types of 3C products employed to compare the rankings of search accuracy, progressive price increase accuracy, and progressive price decrease accuracy, the price comparison system adopted in the study, out of 12 price comparisons, demonstrated eight wins, 2 ties, and 2 losses.
|
|---|
