方志物产挖掘及系统构建  

Produce Mining in the Local Chronicle and Its System Construction

作　　者：衡中青[1,2] 侯汉清[1] 

机构地区：[1]南京农业大学信息科技学院,南京210095 [2]佛山科学技术学院图书馆,佛山528000

出　　处：《情报学报》2009年第2期175-181,共7页Journal of the China Society for Scientific and Technical Information

基　　金：科技部社会公益性项目（2005DIB6J028）研究成果之一;国家社科基金重点项目：《文化典籍整理与开发的智能技术研究》（编号：08ATQ002）研究成果之一。

摘　　要：本文首先根据方志文献的行文特征,设计一个统一规范的方志文献文档处理格式,用于文本文档的预处理。本文建立方志全文数据库,并基于数据库构建一个物产挖掘系统,进行物产信息挖掘,方法如下：通过抽取数据库正名字段内容识别物产正名,通过模式识别方法识别物产异名别称。最后建立包含物产正名和异名别称的物产标引词典,对全部物产文献进行标引,从而挖掘出物产名称。经测试,异名别称的识准率为71.6%,识全率为88.6%,表明系统是可行的。Firstly,this article analyses writing styles of local chronicles.Then it designs a standard form at for the preprocessing of text files.Secondly,based on the fulltext database,a produce mining system was constructed.The process of produce mining was followed：we extracted information from the formal fields to recognize the formal names,and adopted pattern recognition to recognize the alternate names.Finally,we constructed a produce dictionary to index all the produce names.The alternate names recognition test revealed that precision was 71.6%and recall was 88.6%,which told us the system was feasible.

关 键 词：地方志 内容挖掘 模式识别 同义词识别 

分 类 号：G255[文化科学—图书馆学] G250.73