In order to improve the accuracy of errorcorrection for Chinese placename, the cleansing frameworkbased on the address-name index table and bi-directionplacename error correction method are proposed. Firstly,the error correction algorithms in data cleansing area arereviewed. Secondly, the cleansing framework based on theaddress-name index table is proposed for the features ofChinese placename and the low accuracy of existingalgorithms. In this framework, the structure of addressnameindex table and relevant concepts are defined, and anew bi-direction placename error correction method is usedto improve the accuracy of algorithm by using theaddress-name index table. In this method, the concepts ofpartial similarity and whole similarity are introduced, andChinese placename are matched and corrected in forwardfirstly, then they are searched from address-name indextable and corrected in reverse. Finally, extensive simulationexperiments are conducted to prove the feasibility andrationality of method, the results of which show that thebi-direction placename error correction method is betterthan the others in the performance metrics ofimplementation precision ratio and recall ratio for errorcorrection of Chinese placename.
展开▼
