搜索引擎用户满意度多维分析方法的研究

发布时间：2018-11-09 16:22

【摘要】：随着互联网的发展,大数据时代的到来,搜索引擎已经成为了人们获取信息的主要途径。搜索引擎系统一直追求将更好的搜索结果展现给用户,因此,搜索引擎结果质量的评估也成为了搜索引擎厂商关注的重点。搜索引擎厂商根据自己搜索引擎的特点和业务需要,都有一套自己的评估体系,在一次对搜索引擎的改进上线之前都会去评估其好坏。用户满意度是搜索引擎评估中一个十分重要的指标,它直接反映了用户在使用搜索引擎时对搜索引擎返回结果的满意程度。本文就是在这样的背景下,通过分析现有的用户行为点击数据,建立了一套适用于搜索引擎用户满度分析的多维分析方法。首先,对现有的用户点击行为日志进行清洗、转换,分析日志的特点,精选了71个维度属性,建立一个多维数据模型。然后,基于多维数据模型进行多维分析。在进行多维分析时,可能会出现得到的结果难以解释、不够置信,并且由于维度较多,在进行多维交叉分析时,会出现维度爆炸的问题。因此,将关联规则挖掘引入到多维分析方法中,解决结果不够置信以及多维爆炸的问题。最后,通过设计实验,来验证多维分析方法以及关联规则的可行性,同时确定了关联规则挖掘所用的一些参数值。通过本文的研究,一方面为用户满意度的分析建立了一套多维分析方法,使得分析人员能够从多个维度、多个角度去进行用户满意度分析;另一方面,也为以后的多维分析方法设计,以及将关联规则挖掘运用到多维分析方法提供一个研究思路。
[Abstract]:With the development of Internet and the arrival of big data era, search engine has become the main way to get information. Search engine system has always sought to show better search results to users. Therefore, the evaluation of search engine results quality has become the focus of search engine manufacturers. According to their own search engine characteristics and business needs, search engine manufacturers have a set of their own evaluation system, in a search engine will be improved before the launch of the evaluation of its good or bad. User satisfaction is a very important index in the evaluation of search engine. It directly reflects the degree of satisfaction with the result returned by the search engine when using the search engine. In this paper, a set of multi-dimensional analysis methods for user fullness analysis of search engine is established by analyzing the existing user behavior click-through data. Firstly, the existing user click behavior log is cleaned, converted, and the characteristics of the log are analyzed. 71 dimension attributes are selected and a multidimensional data model is established. Then, multidimensional analysis is carried out based on multidimensional data model. In multidimensional analysis, the results may be difficult to explain and can not be trusted, and because of the large number of dimensions, there will be dimension explosion in multidimensional cross-analysis. Therefore, the association rule mining is introduced into the multidimensional analysis method to solve the problem of disbelief in the results and multidimensional explosion. Finally, experiments are designed to verify the feasibility of multidimensional analysis and association rules. At the same time, some parameter values used in mining association rules are determined. Through the research of this paper, on the one hand, a set of multidimensional analysis method is established for the analysis of user satisfaction, which enables analysts to analyze user satisfaction from multiple dimensions and angles. On the other hand, it also provides a research idea for the design of multidimensional analysis method and the application of association rule mining to multidimensional analysis method.
【学位授予单位】：东北师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3

【参考文献】