網路書櫃使用者偏好結構與瀏覽尋書行為之研究

Abstract

本研究以aNobii網路書櫃為研究平台，探討網路書櫃使用者以不同瀏覽管道尋找有興趣的休閒讀物時，其主觀評價與客觀成效表現是否有差異。本研究試圖採用精確性與非精確性概念的評估指標，希望能更貼近休閒讀物尋書之需求。研究中導入行銷學偏好結構的概念，以讀者閱讀的偏好結構作為使用者變項，探討讀者個人偏好結構是否會影響其使用不同推薦工具瀏覽尋書的結果。研究採用準實驗設計，自變項為aNobii上三種導覽工具（朋友書櫃、相似書櫃、同作者書籍）以及受試者個人偏好結構（偏好發展程度、偏好同質性、閱讀涉入程度）；依變項為實驗後問卷搜集的受試者主觀評價，以及在實驗中系統自動於背景搜集的客觀檢索成效。 研究結果發現：一、不同導覽工具的確有所差異。使用者認為同作者書籍較符合其偏好，實際檢索成效亦較佳，但瀏覽起來最無趣；而最不符合偏好的朋友書櫃，卻是三者中瀏覽過程最有趣、亦最能擴展讀者閱讀視野的管道。二、讀者對不同導覽工具的主觀評價結果與客觀檢索成效並不完全一致，但若細就精確性與非精確性的概念區分，可發現主觀與客觀的結果有接近的趨勢。三、讀者的個人偏好結構的確會影響其瀏覽尋書的評價與成效，如偏好發展程度與偏好同質性愈高者，使用同作者書籍瀏覽尋書的客觀成效明顯較佳；閱讀涉入程度愈高者，朋友書櫃中愈容易遇到重覆的已知書籍等等。因此，若此類平台未來能建立讀者的個人偏好結構資訊檔，集有足夠的研究資訊，或許對使用者的服務能更切合所需，也能令使用者的偏好結構與其閱讀需求之間的關係更為清晰。

This study aimed to explore how users search books they are interested in with different browsing tools in online bookshelf, aNobii, and whether the performance in subjective perspective and objective results will be different. To meet the demands of users’ leisure reading, we use accuracy and novelty as evaluation indicators. Besides, applying the concept of preference structure widely used in marketing as our independent variable, we attempted to investigate whether readers’ preference structure will influence results of book selection with different recommendation tools. A quasi-experimental design was adopted where all 40 participants searched alternately with the three book finding tools. There are two independent variables in the research, three browsing tools in aNobii-- friends, similar readers, and the author, and preference structure including preference development, preference homogeneity, and degree of reading involvement. The dependent variables are subjective perspective collected by questionnaires after the experiment and objective search results retrieved by the computer during the experiment. Some major findings are as follows. First, different browsing tools were found to affect users’ book selection behaviors. Users thought that browsing by author could help them find books better matching their reading preference. It produced a higher accuracy. However, users felt it was less interesting when choosing books by authors. On the contrary, users considered browsing friends’ bookshelf to be most interesting. Books found in friends’ bookshelf may not fit their reading preference but helped users broaden their horizon. Second, performance in subjective perspective and objective search results did not align with each other , which suggested they captured different aspects of user experience . Furthermore, readers’ preference structures were found to affect their browsing results. For instances, search results were better when the reader’s preference is highly developed and homogenous. Also, the reader with high reading involvement was found to be more likely to find books they had already known during browsing friends’ bookshelf. Thus, if we can build users’ reading preference profile to collect enough information for research, we can supply service that meet users’ needs in the future. Moreover, it can help us understand the relation between readers’ preference structure and reading needs.