Distributed Cooperative Information Filtering by using Data Mining Techniques

电子数据中的知识发现和数据挖掘的研究具有重要意义，我们正在尝试提出数据挖掘算法，实现对包含书目信息的半结构化数据的网络搜索系统，并开发海量数据的信息可视化接口。在本研究中，我们在之前实现Mondou(RCA.AU)搜索引擎的研究成果的基础上，重点研究了更完善的协作式分布式信息过滤机制。首先，我们描述了一些研究成果，如降低搜索成本的Web数据分析、利用Web超链接图的特征结构提出新的搜索模型、基于P2P网络的协作式分布式信息过滤系统框架。此外，我们还尝试开发基于国家饮食图书馆网络存档系统统计数据的信息系统。在本报告中，我们介绍了网络信息检索系统、数据挖掘工具和信息过滤技术的背景研究问题。接下来，我们描述了我们对几种Web挖掘算法的研究成果，以及基于P2P网络的协作分布式信息过滤系统的部分结果。最后，我们尝试将我们的研究成果推广到网络档案系统的管理和运行中。

Researches of knowledge discovery and data mining from electronic data are important, we are trying to propose data mining algorithms, to implement web search systems for semi-structured data including bibliographic information and to develop information visualizing interface for amount of data.In this research, based on our previous research results of implementing "Mondou (RCA.AU)" search engine, we focus on more sophisticated cooperative distributed information filtering mechanisms. Firstly, we describe several results, such as web data analysis for reducing search costs, propose of new search model by using characteristic structures of web hyperlink graphs, frameworks of cooperative distributed information filtering systems over P2P network. Furthermore, we try to develop information systems based on statistical data of web archiving system of NDL (National Diet Library).In this report, we introduce web information retrieval systems, data mining tools and background research issues of information filtering techniques. Next, we describe our research results of several web mining algorithms, and partial results of cooperative distributed information filtering systems over P2P network. Finally, we try to extend our research results for management and operation of web archiving systems.