Research On Technology For Chinese Address Service

The address with language and words as medium is one form of data which is most close to human,at the same time is an important information carrier to assist the human brain in perceiving the world,which makes a big difference in economy construction and people's living.The information application in different domains is becoming more and more intuitive and convenient with the help of address service.However,the variety of natural language description and differences of human's language habits,make people use different way to describe a same address in the database of different domains and departments.At the same time,due to the intersecting of the government functions,and the relatively backward construction of Chinese address standardization,many problems are exposed in duration of the normative,integrated and consistent construction on the original address data.On this background,while using the address database,different government departments and industry manufacturers usually need to make custom development and develop a particular data service against to owned address data,which makes the spatial data interoperation particularly difficult.Based on researching the existing address service,this paper discuss the progress of address matching with different description rules and the degrees of detailed.Then make a more indepth exploration in the process of address data acquisition,participles,address matching and spatial information reasoning.In the end,we developed a prototype system for Chinese standard address service.In this paper,the main research results include the following main aspects:More suited to the characteristics of the Chinese address segmentation strategy:on the basis of,summarizing the existing Chinese word segmentation and Chinese address segmentation study,analyze the importance of Chinese word segmentation study in Chinese address matching,and design a word segmentation engine which is more suitable for Chinese address characteristics.The Strategy includes collecting and establish a more complete Chinese birthright word dictionary,make an address segmentation character system whose structure is more integrated,with the help of markov model which has achieved well effects in Chinese word segmentation,which will provide a reliable support for the Chinese address matching service behind this.The establishment of a toponymic role recognition engine suitable for Chinese address matching:through the study of the address combination model in the digital city construction and the statistical analysis of the actual target data,Establishing the set of place names for the target data.According to the place name system,the sample data are manually annotated to establish the training set,and then the hidden Markov model would be used to realize the automatic recognition of the place roles.Formulated address matching strategy based on full text index and the character matching:development high matching rate and high accuracy of word segmentation combination model,make an integrated address matching strategy through the character matching model and computing the similarity of participles with characters.Combining with the full text index method which has been widely used in the business product,this address matching strategy will play well both in efficiency and accuracy.Developing standard name address service prototype system:using industrial and commercial address data of Wuhan as the data source,establishing standards address library and the standard library after data cleansing,processing the address data with the help of above strategies,developed the prototype system which.is suitable for industry application.The system will contains Chinese address fuzzy query that performances well both in efficiency and accuracy,standardization against the nonstandard address,address coding/anticoding query service interface and application demonstration,and so on.