网络信息抽取技术及其在TBT预警中的应用*

研究了一种能够实现对数据型网页中信息实施实时采集的信息技术。该技术能够智能识别表格结构，自动分离数据项，在对数据项的分析判断过程中，采用从单词上分类（By Words）和从表格排列方式（By Structure）划分相结合的方法，以Ontology思想为支撑，融合支持向量机算法（SVM）和隐马尔可夫模型（HMM）等一系列成熟模型。最后通过测试并将该技术应用于TBT预警信息动态采集子系统中，收到良好效果。

The Technology of Web Information Extraction and Its Application in the TBT Early-Warning System

This paper researches into an information technology, which could real-timely extract the interested information from data-type Web pages. The technology we employ could intelligently identify table structures, and automatically separate different kinds of data. In the process of analyzing and classifying data, it adopts the combination of sorting by words and dividing by table structure, which depends on the idea of ontology and aggregates a series of mature models, such as SVM and HMM. The technology, which has passed the test, is applied into a dynamic information gathering system of a TBT early-warning system and does a good work.