互动问答平台专家发现及问题推荐机制的研究
发布时间：2018-11-11 20:24
【摘要】：在信息检索领域,与根据用户键入关键字进行检索的搜索引擎相比,互动问答平台使用了语义更加丰富的自然语言。百度知道、Yahoo!Answers、Quora,以及目前人气颇高的知乎,这些互动问答平台已经成为用户获取信息、分享知识的重要渠道。但随着互动问答平台的不断发展,用户数和问答量骤增。对任何一个用户而言,刚提交的问题可能很快就被其他用户新提交的问题给淹没。这种现象带来的后果便是用户提出的问题可能要过很长时间才会有其他用户去回答。与此同时,用户得到的回答可能并不能令其满意,甚至包含了大量垃圾信息。 本文尝试通过对专家发现和问题推荐机制的研究,帮助被动等待的提问者在尽可能短的时间内得到问题的回答,并且这些回答是令其感到满意的。本文首先通过统计方法,分析并总结互动问答平台中的问答情况及其特点。然后,提出了改进的PageRank算法并将其应用到问答社区中的专家发现过程。最后,基于对问答专家发现的研究,设计了互动问答平台的问题推荐架构和推荐流程,旨在针对待解决的问题,系统自动将问题推荐到合适的用户处作答。 作者使用Java代码实现了本文提出的算法,通过实验证明了本文提出的问答专家发现方法的有效性和可行性,并通过基于问题推荐的示例原型系统展示了问题推荐的流程。
[Abstract]:In the field of information retrieval, the interactive question and answer platform uses a more semantic natural language than the search engine which searches according to the key words typed by the user. Baidu knows that Yahoo AnswersQuora, and the current popularity of these interactive Q & A platforms have become an important channel for users to get information and share knowledge. However, with the continuous development of interactive Q & A platform, the number of users and the number of Q & A have increased. For any user, a newly submitted question may soon be overwhelmed by a new one submitted by another user. The consequence is that it may take a long time for other users to answer questions. At the same time, users may not be satisfied with the answer, and even contain a lot of spam. This paper attempts to help the passive questioner get the answer to the question in the shortest possible time by studying the mechanism of expert discovery and question recommendation, and these answers are satisfactory to him. This paper firstly analyzes and summarizes the Q & A and its characteristics in the interactive Q & A platform by means of statistical method. Then, an improved PageRank algorithm is proposed and applied to the expert discovery process in the Q & A community. Finally, based on the research of question and answer experts, the question recommendation framework and process of interactive question answering platform are designed, aiming at solving the problems, the system automatically recommends the questions to the appropriate users to answer. The author uses Java code to realize the algorithm proposed in this paper. The experiment proves the validity and feasibility of the method of question and answer expert discovery, and shows the flow of problem recommendation through an example prototype system based on question recommendation.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.3
[Abstract]:In the field of information retrieval, the interactive question and answer platform uses a more semantic natural language than the search engine which searches according to the key words typed by the user. Baidu knows that Yahoo AnswersQuora, and the current popularity of these interactive Q & A platforms have become an important channel for users to get information and share knowledge. However, with the continuous development of interactive Q & A platform, the number of users and the number of Q & A have increased. For any user, a newly submitted question may soon be overwhelmed by a new one submitted by another user. The consequence is that it may take a long time for other users to answer questions. At the same time, users may not be satisfied with the answer, and even contain a lot of spam. This paper attempts to help the passive questioner get the answer to the question in the shortest possible time by studying the mechanism of expert discovery and question recommendation, and these answers are satisfactory to him. This paper firstly analyzes and summarizes the Q & A and its characteristics in the interactive Q & A platform by means of statistical method. Then, an improved PageRank algorithm is proposed and applied to the expert discovery process in the Q & A community. Finally, based on the research of question and answer experts, the question recommendation framework and process of interactive question answering platform are designed, aiming at solving the problems, the system automatically recommends the questions to the appropriate users to answer. The author uses Java code to realize the algorithm proposed in this paper. The experiment proves the validity and feasibility of the method of question and answer expert discovery, and shows the flow of problem recommendation through an example prototype system based on question recommendation.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.3
【参考文献】
相关期刊论文 前4条
1 邢春晓;高凤荣;战思南;周立柱;;适应用户兴趣变化的协同过滤推荐算法[J];计算机研究与发展;2007年02期
2 费洪晓;蒋，
本文编号：2326045
本文链接：https://www.wllwen.com/kejilunwen/sousuoyinqinglunwen/2326045.html
