Research On Character Recognition And Fusion Method Of Internet

Nowadays,the internet has become the necessities of people's life and work,billions of people get or release information on the Internet every day,so the Internet has become the ocean of information.The information on the Internet is complex and diverse,and in this modern society,people pay more attention to human beings,so person information become the focus of the researchers.The geographic location attribute has the most extensive application among all person information attributes,for example location information is widely used in advertisement recommendation,event locating and character behavior research,Therefore,the identification of the persons' geographical location has also become a hot spot in the present research.Among all kinds of network types of internet,the social networks have attracted wide attention because of their huge amount of users and the real-time characteristics of interaction,and the social networks contain the most abundant person information because of their characteristics.So this thesis mainly does the research of persons' location identification about social networks,and there are two categories of person location attributes,called mentioned location and home location,and user's home location refers to the main work and living area of the user.The main research and innovation of these two kinds of location attributes are as follows:(1)A location entity recognition method based on the geographical features and feature bagging is proposed.This method is used to identify the mentioned location in the user's tweets.The method mainly improves the shortcomings of the traditional named entity recognition method to obtain the location entity,and the method provides a rich and targeted machine learning features for the place name entity,and those features includes individual features and combination features.It is proved by experiments that these features can effectively improve the recognition effect of the model.At the same time,in order to solve the problem of features weight undertraining in the traditional model,this method uses feature bagging,this method solves the problem and improves the stability of the model,and also improves the recognition effect.(2)A method of user home location recognition based on multiple information fusion is proposed.This method uses a variety of information to solve the existing problem that many researchers do this research with poor means,those information mainly includes the user's social network friend information and the user's self-information,and self-information includes user's tweet information,user account and description information,and user's following list.The fusion of information by means of unified information representation and geographic location clustering,the home location of the user is identified by the weighted calculation of the latitude and longitude of the cluster position.In this thesis,we find that the recognition method about multiple information can effectively identify the home location of the user.