基于规则的百科人物属性抽取

Rules-Based Character Attributes Extraction from Baidu Encyclopedia

信息抽取是数据挖掘的一个重要领域,文本信息抽取是指从一段自由文本中抽取出指定的信息并将其结构化数据存入知识库供用户查询或下一步处理所用。人物属性信息抽取是智能人物类搜索引擎构建的重要基础,同时结构化信息也是计算机所能理解的一种数据格式。作者提出了一种自动获取百科人物属性的方法,该方法利用各属性值的词性信息来定位到百科自由文本中,通过统计的方法发现规则,再根据规则匹配从百科文本中获取人物属性信息。实验表明该方法从百科文本中抽取人物属性信息是有效的。抽取的结果可以用来构建人物属性知识库。

Information extraction is an important area of data mining. Text information extraction means extracting specified information from a section of free text and storing structured data in the knowledge base for user querying or further processing. Character attribute information extraction is an important instrument of building search engine of persons, and is also a technology for computer program understanding. This paper presents an automatic method to obtain encyclopedia character attributes, and this method uses the speech tagging of each attribute value to locate the encyclopedia free text. The rules are discovered by statistical method, and the character attributes information is obtained from encyclopedia text according to rules matching. Experiments show that this method is effective in extracting character attribute information from encyclopedia text. The extracted results can be used to build the knowledge base of the character attributes.