一种基于规则的模糊中文地址分词匹配方法  

A Rule-Based Segmenting and Matching Method for Fuzzy Chinese Addresses

作　　者：程昌秀[1] 于滨[1] 

机构地区：[1]中国科学院地理科学与资源研究所,北京100101

出　　处：《地理与地理信息科学》2011年第3期26-29,共4页Geography and Geo-Information Science

基　　金：国家863项目"经济普查与基本单位统计遥感应用系统"(2006AA120106);"地理空间数据库管理系统总体设计"(2007AA120401)

摘　　要：在研究分析地址模型的基础上,建立了存储标准地址数据集的标准地址库和自定义的地址匹配规则库,提出了一种基于规则的模糊中文地址编码方法。该方法在依据标准地址库分词的同时,也沿着自定义的地址匹配规则进行推理,从而缩小了下次分词所用到的目标数据集,提高了系统执行效率。另外,通过借助构建的规则树与歧义栈,提高了文中定义的两类模糊地址匹配的成功率。最后,基于该算法建立了一个地理编码原型系统,并利用经济普查项目中的相关数据对算法的可用性进行了验证。After analyzing Chinese address model,this paper built a standard address database and an address matching rules database,and then presented a rule-based Geocoding method for fuzzy Chinese addresses.This method used the standard address database to segment the input fuzzy Chinese address.At the same time,the method used the rules database to reduce and find a standard address that matched with that fuzzy address.The method used the customized rules to reduce candidate addresses so that it can participate in match reduction and save the matching executive time.In addition,the introduction of rule tree and semantic stacks also promote the matching of fuzzy address.Finally,a Geocoding prototype system was built,and then its availability was verified utilizing the data of natural economic census project.

关 键 词：地理编码 模糊地址 规则库 地址分词 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]