用户评论的质量检测与控制研究综述

Survey on Quality Evaluation and Control of Online Reviews

随着网络技术的发展,越来越多用户生成的内容(user-generated content)出现在网络应用中,其中,用户评论富含用户的观点,它们在网络环境中充当越来越重要的角色.据美国Cone公司2011年的调查报告,64%的用户在购买行为之前会参考已有的用户评论.因此,为用户提供准确、简洁和真实的评论是一个迫切且重要的任务.主要围绕评论质量评估、评论总结和垃圾评论检测这3个方面综述了国际上评论质量检测与控制的研究内容、技术和方法的研究进展.在此基础上,展望该领域的发展给出了可能的研究方向.

With the development of Web2.0, more and more user-generated content （UGC） occur in Web applications. These contents, especially reviews, are opinion-rich and play important roles in e-commerce. According to the Cone＇s survey published in 2011, 64% of users like to read the related reviews on goods before they make purchase decisions. It is vital to provide users the accurate, succinct and true reviews. This work surveys the research progresses on quality evaluation and control of reviews focusing on the prediction, summarization and spam review detection. At the end, some potential research topics are pointed out based on these analyses.