| Summary: | 碩士 === 國立臺灣大學 === 圖書資訊學研究所 === 107 === A study was conducted to compare the book finding performance between social (i.e. customers who bought this also bought) vs. subject based navigation on books.com.tw. To fully capture the effectiveness of the exploratory-based book finding tools, three aspects of evaluation were proposed, namely, user experience, search precision, and the quality of the search result. Furthermore, besides traditional accuracy-based criteria, search novelty and serendipity were also introduced in response to recent calls for “non-obviousness” measure that better reflect consumer value when interacting with recommender system. A Latin square experiment was adopted where 52 participants searched fictional vs. non-fictional books alternatively with two book finding tools. Another object of the study was to test the moderating role of users’ preference characteristics, including “preference diversity”, “openness to novelty,” “involvement” and “preference insight” on search performance.
It was found that social navigational tool produced a large consideration set, which was also preferred for future use. Though the social navigation tool generated less novelty and serendipity but higher decision confidence when searching for fictions, mainly because the current algorithm on books.com.tw tend to bring up popular sellers. Furthermore, the social navigational tool is particularly effective in terms of judgment confidence for users with high preference diversity searching for non-fictions, indicating that the social navigation tool was able to bring up more diverse results in non-fictional books.
|
|---|
