Research And Application On Recommender Method For Ranking Lawyers Based On Text Mining

With the development of legalization process in China, it become more and more open for trial of cases, for example, the appearance of written judgement on website. While people realized that it is a better way to choose legal weapon to resolve dispute, lawyers become more and more important. With the development of the Internet technology, lots of information about lawyers appear on the Internet. However, existing technology can not offer a effective way to rank lawyers. Faced with a great deal of information, people realized that it is hard for them to find a suitable lawyer as soon as possible.To solve problems mentioned above, this paper offers a novel method for recommend lawyers based on technology of text mining and recommender system. This method uses text mining technology to extract information from written judgement which published on the website of local court, and these information are used to rank lawyers. Base on user record and the structure of written judgement, lawyers can be recommended. This method contains following aspects:First of all, a crawler program base on static template should be design to obtain written judgement. Text classify method will be used to classify written judgement which missed its class information, and to meet query request from users, this paper build a full-text index for amount of written judgement.Then, based on information extraction technology and structure of written judgement, this paper designed a method to extract information from written judgement using rule matching. The written judgement can be classified to civil, criminal and administrative, and each class has different information to extract. So, this paper designed different rules for each class of written judgement.Finally, Using information from information extraction, lawyers can be ranked. These information can’t be used for computation because most of them are text information. So, first of all, text information should be transformed to numerical value, and each information will take different threshold to distinguish there importance. Finally, by sorting the score of each lawyer, we can get the result of lawyers ranking. Based on relationship in written judgement, lawyers can be recommended by using the rank result and users’ record. Accuracy and feasibility are testified by compared with traditional method.