| 基于HowNet的话题跟踪及倾向性分类研究 |
| |
| 引用本文： | 金珠,林鸿飞,赵晶.基于HowNet的话题跟踪及倾向性分类研究[J].情报学报,2005,24(5):555-561. |
| |
| 作者姓名： | 金珠 林鸿飞 赵晶 |
| |
| 作者单位： | 大连理工大学计算机科学与工程系,大连,116024 |
| |
| 基金项目： | 国家自然科学基金资助项目(60373095) |
| |
| 摘 要： | 本文研究了如何基于信息检索技术和“知网”实现有效的话题跟踪和话题立场分类。话题跟踪任务就是给出话题相关的训练新闻报道,系统在后续报道中发现与这个话题相关的报道。它属于话题检测与跟踪的一项子任务。本文针对跟踪任务中话题本身的特点,论述了权重调整、事件框架和报道扩充等多种提高跟踪性能的策略,同时基于“知网”中的情感体系和动态角色框架,提出了如何填充框架并结合建立的立场概念库对报道进行话题立场分类。实验证明这些方法是有效的。
|
| 关 键 词： | 话题检测与跟踪 信息检索 知网 向量空间模型 |
| 修稿时间： | 2004年11月4日 |
Study on Topic Tracking and Tendency Classification Based on HowNet |
| |
| Jin Zhu,Lin Hongfei,Zhao Jing.Study on Topic Tracking and Tendency Classification Based on HowNet[J].Journal of the China Society for Scientific andTechnical Information,2005,24(5):555-561. |
| |
| Authors: | Jin Zhu Lin Hongfei Zhao Jing |
| |
| Abstract: | This paper describes research into the traditional IR technique to build effective Topic Tracking systems and topic tendency classification based on HowNet. Topic tracking involves tracking a given news event in a stream of news stories i.e. finding all subsequent stories in the news stream that discuss the given event. This research has grown out of the Topic Detection and Tracking (TDT) initiative sponsored by DARPA. Paper focuses on the point of topic and discusses the methods of weight adjustment, event frame and story expansion to improve the tracking effectiveness. Finally paper discusses classification of story tendency based on affective words and event role frame in HowNet. It has been verified by our experiments. |
| |
| Keywords: | topic detection and tracking information retrieval HowNet vector space model |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
