网络信息构造与用户行为结合分析研究

发布时间：2018-11-07 08:58

【摘要】：伴随着人们越来越多样化的信息需求,问答社区(Community Question Answering, CQA)网站服务模式应运而生。广泛的用户参与使其信息量迅速增长,庞大的信息资源库也为搜索引擎提供了很好的信息源,越来越多Web用户通过搜索从中获取信息。然而,信息的长期积累造成大量“过时”信息出现在其中,给使用者带来不便。网页浏览日志记录着用户浏览过程中的细节信息,反映用户行为、意图和使用习惯,对分析CQA查询用户使用情况和信息时效性有着重要意义。本文提出针对用户网页浏览日志的处理方法,包括URL查询关键字的截取与规格化处理、查询过程的划分等。在查询过程划分的基础上,对大量真实用户的浏览行为习惯做了统计分析。结果显示,用户每查询一次信息平均用时6.28分钟、访问8个网页；部分查询在交替并发中进行；用户对于各网站站内搜索引擎使用频率较高。本文结合用户浏览行为分析,以及CQA信息固有特征,建立CQA查询用户满意度判断框架。结果表明在加入用户浏览行为特征后,分类框架的准确率、召回率均有明显提升。通过分析Yahoo Chiebukuro问答社区的用户满意率和信息时效性,发现用户满意率和信息时效性在不同问题类别之间的表现差异明显。
[Abstract]:Along with people's more and more diversified information demand, the question and answer community (Community Question Answering, CQA) website service pattern arises at the historic moment. The extensive user participation makes its information quantity increase rapidly, the huge information resource database also provides the very good information source for the search engine, more and more Web user obtains the information through the search. However, the long-term accumulation of information causes a lot of "outdated" information to appear in it, causing inconvenience to users. Web browsing log records the details of the user's browsing process and reflects the user's behavior, intention and usage habits. It is of great significance to analyze the usage of CQA query users and the timeliness of information. In this paper, the processing methods for user's web browsing log are proposed, including the interception and normalization of URL query keywords, the partition of query process and so on. Based on the partition of query process, the browsing behavior of a large number of real users is analyzed statistically. The results show that the average time of each query is 6.28 minutes, and 8 pages are visited; part of the query is carried out alternately and concurrency; the users use the search engine in each site with a high frequency. Based on the analysis of user browsing behavior and the inherent characteristics of CQA information, this paper establishes a framework for judging the satisfaction of CQA query users. The results show that the accuracy and recall rate of the classification framework are improved obviously after the user browsing behavior feature is added. By analyzing the user satisfaction rate and information timeliness of Yahoo Chiebukuro Q & A community, it is found that the performance of user satisfaction rate and information timeliness in different problem categories is obvious.
【学位授予单位】：北京化工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092

【参考文献】