高级检索|图表检索

图书情报工作 ›› 2019, Vol. 63 ›› Issue (22): 106-117.DOI: 10.13266/j.issn.0252-3116.2019.22.012

• 情报研究 • 上一篇    下一篇

基于在线评论的图书消费者满意度影响因素与作用机理

尹丽春1,2, 王悦1   

  1. 1. 黑龙江八一农垦大学经济管理学院 大庆 163319;
    2. 山东青年政治学院政治与公共管理学院 济南 250000
  • 收稿日期:2019-04-28 修回日期:2019-07-02 出版日期:2019-11-20 发布日期:2019-11-20
  • 作者简介:尹丽春(ORCID:0000-0002-5510-1349),教授,博士,E-mail:308895957@qq.com;王悦(ORCID:0000-0003-4032-0263),硕士研究生。
  • 基金资助:
    本文系黑龙江农垦总局基金项目"大数据时代黑龙江垦区科技创新战略研究"（项目编号：HNK125B-14-08A）和黑龙江八一农垦大学研究生创新科研项目（项目编号：YJSCX2018-Y76）研究成果之一。

Influencing Factors and Mechanism of Book Consumer Satisfaction Based on Online Comments

Yin Lichun1,2, Wang Yue1   

  1. 1. School of Economics and Management, Heilongjiang Bayi Agricultural University, Daqing 163319;
    2. School of Politics and Public Administration, Shandong Youth University of Political Science, Jinan 250000
  • Received:2019-04-28 Revised:2019-07-02 Online:2019-11-20 Published:2019-11-20

PDF

摘要： [目的/意义] 本文旨在提出一种从大量在线商品评论数据中挖掘影响读者满意度的关键因素的方法，并深入探讨各个影响因素对消费者满意度的影响模式和影响程度，进而为图书出版企业、电商平台持续改善读者满意度提供理论基础。[方法/过程] 一方面利用朴素贝叶斯分类器将读者的情感进行分类。另一方面对评价文本中的高频名词进行聚类，发现影响读者满意度的主要因素。在此基础上基于最大程度减少不确定性的原则对各个影响因素的影响模式和影响程度进行分析。[结果/结论] 以京东人工智能类图书评论为例进行了实证研究，发现包括内容、价格在内的六个因素可以极大地反映出读者的满意度（83.2%）。因此对于图书类商品，可以通过对大量历史评论数据的学习，找出影响读者满意度的主要因素，据此设计出简化的读者评论框架，以增强读者参与评论的积极性，提高评论的质量。"图书内容"是影响读者满意度的最主要因素。当读者对图书内容表达出不同的情感时，其他因素对读者满意度的影响模式和程度是完全不同的。当读者对图书内容表示满意时，89.2%的总体评论是好评，其他因素的影响较小，价格是导致中评和差评的最主要因素；当读者认为图书内容一般时，评论趋向于中性，读者对服务和物流更为关注；当读者认为图书内容不令人满意时，影响读者满意的因素依次是包装、服务质量和价格。基于不同影响因素对读者满意度的影响模式和影响程度，图书出版企业和电商平台可以更加有针对性地对其加以改善，以提高读者满意度。

关键词: 读者满意度, 图书在线评论, 机器学习, 情感分析

Abstract: [Purpose/significance] This paper aims to propose a method to mine the key factors influencing reader satisfaction from a large amount of online commodity review data, and deeply explore the influence mode and degree of each influencing factor on consumer satisfaction, so as to provide a theoretical basis for book publishing enterprises and e-commerce platforms to continuously improve reader satisfaction.[Method/process] On the one hand, naive bayesian classifier was used to classify reader emotion; On the other hand, the high frequency nouns in the evaluation text were clustered to find the main factors that affect reader satisfaction. On this basis, the influence mode and degree of each influencing factor were analyzed based on the principle of minimizing uncertainty.[Result/conclusion] Taking the JD's artificial intelligence book review as an example, an empirical study was conducted. It was found that the six factors including content and price can greatly reflect the reader's satisfaction (83.2%). Therefore, for books and commodities, we can find out the main factors affecting readers' satisfaction by studying a large number of historical review data, and design a simplified reader review framework accordingly, so as to enhance the enthusiasm of readers to participate in the review and improve the quality of the review. ‘Book content’ is the most important factor affecting reader satisfaction. When readers express different emotions on the content of the book, the influence mode and degree of other factors on reader satisfaction are completely different. When readers are satisfied with the content of the book, 89.2% of the overall comments are favorable, while other factors have little influence. When readers think the book content is general, the review tends to be neutral, and readers pay more attention to the service and logistics; When readers think the content of books is not satisfactory, the factors affecting the satisfaction of readers are packaging, service quality and price in order. Based on the influence mode and influence degree of different influencing factors on reader satisfaction, book publishing companies and e-commerce platforms can be improved more specifically to improve reader satisfaction.

Key words: reader satisfaction, book reviews online, machine learning, sentiment analysis

中图分类号: