基于规则的中文地址要素解析方法  

Rule-based Approach to Semantic Resolution of Chinese Addresses

作　　者：张雪英[1] 闾国年[1] 李伯秋[1] 陈文君[1] 

机构地区：[1]南京师范大学虚拟地理环境教育部重点实验室,南京210046

出　　处：《地球信息科学》2010年第1期9-16,共8页Geo-information Science

基　　金："863"项目(2007AA12Z221);国家自然科学基金项目(40971231);南京师范大学重点科研基金资助项目(2006105XGQ0051)

摘　　要：在日常生产与生活中,地址是最常见的使用自然语言描述地理位置的参考系统之一。地址地理编码技术被认为是大量业务数据的G IS实现可视化定位和空间分析的桥梁,在房地产管理、土地管理、城市规划、公安、邮政、税收、电讯和公共卫生等领域中具有十分重要的应用前景。地址要素解析是中文地址编码的核心技术之一。它是将自然语言描述的地址拆分为在某一限定区域内,可以指定某一地理范围的地址要素的过程。实际上,这个过程可以被看作是一种特定的中文分词任务。由于语言和文化的原因,中文地址描述采用连续的字符串,而且普遍存在不规范现象。目前,采用的地址解析方法在较大程度上受限于词典的更新维护和规则的不完备问题。本文以大规模地名词典和地址数据库为数据源,通过系统分析地址要素的构词特征和句法模式,构建了各类地址要素的特征字库,提出了中文地址的数字表达方法,设计了RBAI中文地址要素解析算法,开发了相应的原型系统。实验结果准确率达到92%以上,处理效率达2800条/秒。这表明该方法符合大规模数据处理的应用需求,具有重要的推广应用价值。A geographic information system （GIS） integrates hardware, software, and data for capturing, managing, analyzing, and displaying all forms of geographically referenced information. Addresses are one of the most popular geographical reference systems in natural languages. Address geocoding is considered as the most effective approach to bridging the gap between business data in management information systems （MIS） and GIS, which supports geospatial information visualization and spatial analysis. Chinese address geocoding faces three significant problems, i.e. address models, address resolution and address matching, because of the un-standardization of Chinese place names and the shortage of national address databases. Address resolution aims to automatically split address strings in natural language into address units without semantic incompletion. It plays a fundamental role in address models and address matching. Previous research focuses on rule or gazetteer based approaches, which are easily implemented but with poor coverage and performance. In theory, Chinese address resolution is similar to word segmentation in Chinese natural language processing. Based on the investigation of large-scale Chinese place names and address syntactic patterns, this paper identifies primary and secondary general characters that represent a variety of address units. And then an address numerical representation method is presented to induce syntactical rules of Chinese addresses. Finally, we develop an RBAI algorithm for implementation Chinese address resolution and illustrate an example. The experimental results indicate that the proposed approach can achieve satisfactory efficiency and effectiveness for large-scale data processing, the accuracy ratio over 92% and the processing rate over 2 ,800 items per second. The proposed approach and system can be extended to such fields as land management, asset management, city plan, public security, postal system, taxation, public health management and other location-base services

关 键 词：中文地址 语义解析 地址编码 地址表示 

分 类 号：P208[天文地球—地图制图学与地理信息工程]