面向市场情报的Web实体事件发现与踪迹分析研究

项目名称： 面向市场情报的Web实体事件发现与踪迹分析研究

项目编号： No.61303005

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 闫中敏

作者单位： 山东大学

项目金额： 23万元

中文摘要： 实体踪迹隐藏在Web上孤立离散的海量页面中，不能被直接获取利用。针对Web海量信息，建立Web实体事件发现和踪迹分析体系，自动识别实体发展变化中有价值事件以及事件间关系并有机地组织起来，发现Web实体踪迹用以趋势预测和决策支持，成为非常迫切而重要的任务。本项目将建立一套比较完善的理论研究和应用方法，针对市场情报领域实现1）设计实体事件获取模型，在海量Web信息中发现有价值事件并融合，全面准确地获取目标实体事件信息；2）识别实体事件之间的因果关系、从属关系、跟随关系等，按照实体事件的发生时间、发生地点、行为主题、相互关系等不同维度将实体事件信息组织在一起，形成实体踪迹图模型；3）根据实体踪迹图模型整理和提炼实体的周期性踪迹和典型踪迹，进行实体踪迹的多维分析，发现实体发展规律，对外提供深层次的实体信息服务。在未来的几年中该项研究在市场情报分析、商业竞争情报等方面都有着广阔的发展空间和应用前景。

中文关键词： 实体踪迹；事件识别；事件融合；踪迹分析；关联关系识别

英文摘要： There is an abundance of valuable information of entities On Web, but the applications of these information are really superficial. It is very urgent and important to construct an event discovery and trace analysis system of Web entity for Market Intelligence,which aims at processing the large amount of Web information, as well as identifying valuable events and event relation automatically and integrating them organically, then finding the rules of entity evolution and supporting decision-making and trend projections. It will establish a set of relatively improved theoretical research and application methods, in order to finish the targets as follows: First, we will design entity valuable events classing model to identify and fuse valuable event in the vast amounts of Web information, and thus we can get target entity event information accurately. Second, identify event relation, such as causality ,part of、follow, then organizes the entity event information according to the commercial activities, topics, related products, occurrence time, locations, event relation and other different dimensionality of the same entity. Third, we will manage and refine entities evolutionary trajectory by distinguishing the causes, effects and the influence among entity events depending on the entity-tracing of graph model, which

英文关键词： entity trace；entity events identifying；event fusion；trace analysis；relation identifying