基于搜索反馈的移动用户个性化要素型事件摘要模型研究

项目名称： 基于搜索反馈的移动用户个性化要素型事件摘要模型研究

项目编号： No.61472290

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 计算机科学学科

项目作者： 文卫东

作者单位： 武汉大学

项目金额： 80万元

中文摘要： 移动用户个性化事件摘要是社会计算、信息抽取以及自然语言处理领域中一个重要研究课题，移动用户个性化事件摘要是以移动用户为中心的概要化事件知识表示，它能被搜索引擎用以更加复杂的智能搜索任务。当前系统地开展移动用户个性化事件摘要的研究工作较少。据此，本项目提出了使用搜索引擎反馈过滤、用户已有知识体系分析以及提出要素化的事件模型去进行移动用户个性化事件摘要构建模型的研究课题。具体地，本课题包括以下几个主要方面：（1）提出一个移动用户分析模型，可以分析出移动用户兴趣和已有知识分布，可用其作为搜索反馈过滤的依据；（2）提出基于搜索反馈的奖惩型内容过滤方法，可以移动用户兴趣和已有知识分布给予搜索引擎反馈内容分别进行过滤后；（3）提出要素型的事件模型以及基于该模型的事件摘要系列研究以及方法，该方法通过该构建模型自动从过滤结果中抽取事件要素的模板以及事件要点，进行事件模型填充，组合成用户满意事件摘要。

中文关键词： 事件抽取；用户分析；搜索过滤

英文摘要： Mobile User Personalized Event Summarization is an important research topic in the cross field of Social Computing, Information Retrieval and Natural Language Processing. Mobile User Personalized Event Summarization is the wraparound representation of event knowledge centered on mobile user and must will be utilized to process Smart Search tasks by Search Engine in the near future. However, at present, there is little research on performing systemically personalized event summarization of mobile user, which motivates us to propose this research project that exploit search feedback filter, user-obtained knowledge analysis and element-based event model on studying personalized event summarization model of mobile user. Specifically, the research project includes the following parts: (1) we propose a mobile user analysis model which is used to obtain mobile user interest distribution and obtained knowledge distribution, and the results is the basis of search feedback filtering; (2) we present incentive-based filter based on search feedback, which uses the user interest distribution( rewarding ) and user obtained knowledge distribution( punishing ) to filter user-irrelevant search feedback content; (3) we propose a new element-based event model and a series of study and of event summarization based on this element-based event model, which can automatically extract event element templates and event gists to actively complete event model, and organize a complete and user-intent-matching event summary.

英文关键词： Event Extraction;User Analysis;Search Filtering