Emergent Topic Detection Technology Research In Micro-blog Based On Geographical Location Information

With the further development of mobile Internet, Location Based Service(LBS) is getting close to people’s lives. So the number of microblogs carrying geographic location information is increasing in the microblogging space. It becomes a new direction in the social networking burst topic detection areas that how to detect the burst topic from the complex Microblogging space combined with the geographic location information. Researches find the social network burst topic detection technology combined with LBS is still in its infancy currently. Based on existing research, this paper focuses on the burst topic detection technology combined with geographic location information.Because microblog data acquisition technology is complex and inefficient, this paper proposes a Chinese microblogging data collection method based on a set of scheduling applications and existing microblog data acquisition technology. The method can break the restrict of visiting number of API interfaces on the microblogging platform effectively, so it can make the microblog data collection more efficient and simple.Because current Chinese microblogging burst topic detection methods don’t make full use of the geographic location information that microblogging platform generated in the mobile Internet environment, this paper defines the "degree of participation" concept based on the in-depth study of microblog user relationship, the word weight, the topic of user influence and burst detection model. The concept represents the relationship between the Wiebe user and the topic that the Wiebe user’s tweets described. And it is used for improving the user’s influence who has a strong "degree of participation". This paper presents a method that calculate the term weight combined with the geographic location information. The method is used for improving the extraction accuracy of the feature word. So it can improve accuracy of burst topic detection. In the implementation process of agglomeration clustering algorithm,this paper presents an optimization strategy to improve the clustering algorithm efficiency.Finally, the experimental results showed the effectiveness of the proposed method, and the method provides some new perspectives for the researches of the burst topic detection.