Research And Implement Of Named Entity Recognition Technology In The Microblog Conservation Thread

The application platform of Microblog has a huge number of users,and the interaction between them are always real-time.Thus there is a lot of valuable information inevitably in microblogging.But some unique characteristics of microblog led to the existing named entity recognition methods can not reach the desired results.Therefore,we need make a specialized named entity recognition tool for microblogging text.In this paper,on the basis of the existing research for Chinese named entity recognition technology,I mainly have done some works about the following two aspects:One is building microblogging conservation thread.The works in this part is mainly aimed to solve this problem that the microblogging text can not provide sufficient information when extracting the named entities.To solve this problem,I proposed a method that is merging the bowen and its comment to construct the microblogging conservation thread.Namely to use the comment of each bowen to increase the length of the each bowen.The other is using the roles of Chinese name,the roles of Chinese place,the roles of Chinese organization in the process of extracting the named entity.The works in this part is mainly aimed to solve this problem that the informal syntax of microblogging text affects the effectiveness of named entity recognition.I proposed a solution: Adding names,places and the organization roles in the process of extracting named entity.We can not only use grammar rules when recongining the Named entity,but also can take advantage of the semantic features of the context of the named entity.Tthe solution above I proposed is just to use the latter point.In this paper,I have verifyed the method of Named Entity Recognition Technology In Microblog Conservation Thread on the dataset of Sina microblog,and the precision,recall and F measure are 83.5%?77.3% and 80.3%.so the experimental results show that the method can improve the NER effects.