摘要
本文设计并实现了一个基于内容信息过滤的智能Agent:CuteSearcher.它能够根据用户提交的示例文档,采用机器学习的方法对用户的兴趣进行建模,并通过与WWW上的搜索引擎相互作用,自动查找用户所需的信息.在两个方面解决了现有过滤系统在学习方法方面的不足:第一,采用了概念推理网解决词组分割问题;第二,使用基于协同演化的遗传算法解决文档的特征抽取问题.测试表明它可以有效地帮助用户在WWW上搜索信息.
The paper present an information-filtering Agent: CuteSearcher. It can learn the user's information need from sample documents that the user classifies as relevant or irrelevant to his interests, and generate a model for the user in finding new interesting documents. CuteSearcher extends the state of the art in Web-based information retrieval in two ways. First, it uses concept inference network to find word tulles to represent documents. Second, CuteSearcher uses cooperative evolution GA to select features to represent documents. The performance of the system was experimentally evaluated, which shows that CuteSearcher can help users to effectively browse and search the web.
出处
《南开大学学报（自然科学版）》
CAS
CSCD
北大核心
2004年第2期82-87,共6页
Acta Scientiarum Naturalium Universitatis Nankaiensis
基金
天津市自然科学基金资助项目(023601411)
关键词
信息过滤系统
智能AGENT
协同演化
遗传算法
概念推理网
机器学习
information filtering systems ( information filtering Agent
cooperative evolution
genetic algorithm
concept inference network
