Technology Integration for Large-scale Knowledge Processing

研究了搜索引擎和数据挖掘技术的集成及其在海量、异质基因组数据库中的应用。首先，我们尝试基于搜索引擎中存储的索引信息对基因组数据库中的文本进行自动聚类。然而，“词”的索引不足以进行精确的分类，因为在基因组数据库等科学文本数据中，技术“术语”的频率是必不可少的。然后，我们从基因组数据库中构建了一种本体(即技术术语的受控集合)。最后，通过引入关联规则挖掘算法，基于链接信息和本体这两种数据条目之间的关系，对我们的基因组数据库搜索引擎进行了功能上的增强。在针对基因组数据库的迭代搜索和检索期间，我们的系统可以显示用户感兴趣的一组数据条目(即搜索结果)的公共和特定的链接和语言信息。这项服务是通过WWW提供的，因为我们系统的响应足够快。

We studied on the integration of technologies about search engine and data mining and its application to vast amount and heterogeneous genome databases. First, we tried automatic clustering of texts in genome databases based on the index information stored in search engine. However, indices for "words" were not sufficient for precise clustering since the frequency of technical "terms" is essential in scientific text data like genome databases. Then, we constructed a sort of ontology (i.e. controlled collection of technical terms) from genome databases. Finally, we made functional enhancement to our search engine for genome databases by introducing a data mining algorithm called association rule mining, based on the two kinds of relationships among data entries, i.e, link information and ontology. Our system can show link and language information common and specific to a set of data entries (i.e. search result), in which a user is interested, during the iterative search and retrieval against genome databases. This service is provided via WWW, since the response of our system is sufficiently quick.