Research And Implementation Of Topic News Analysis System

News Analysis research is a cross-frontier disciplines with a wide range of applications,its in-depth research will have greatly contributed to the news reports information organization technologies as well as traditional data mining technology.Also,it plays an active role in the fields of information retrieval,decision support,and text mining applications.Internet media has a lot of press releases,through the study of the subject,people can be a good understanding of the attitude of the press releases in the dossier held by topic,the topic of public concern heat,people in different parts of the degree of involvement of the topic,as well as problems involved in the topic from different angles,the relationship between the news and other topics are to keep abreast of the people and an important window to observe public opinion,the development of relevant policy making departments to provide a reference,has a very important practical value and social benefits.Research of this paper is to study and to do jointly with Xinhua News Agency,we will build with natural language processing and data mining technology for people,places,institutions,an important part of keyword extraction and sorting;aid sentiment analysis technology,complete emotional topics analysis;positive and negative news analysis and topic-based news analysis time of heat;manuscripts quantity distribution analysis Timeline and build special news text collection of information visualization and presentation model the conceptual model of the key elements for visual show.The main research work of this thesis is the following four points:Firstly,the research and implementation of news sentiment analysis algorithm.This paper analyzes the role of emotion in the press level.Sentiment analysis of text news can understand the attitude of media and experts in the field some event or state policy.Masses make their own judgments convenient.For news sentiment analysis algorithm research,this paper utilize natural language processing,data mining algorithms.Secondly,the research and implementation of entity extraction algorithm.In this paper,we utilize CRF algorithm in the domain of extraction entities.By training the model automatically recognizes the names,places and organizations.For the entity of time and date,this paper uses the methods of template matching.Thirdly,the research and implementation of keyword extraction algorithm.This paper utilizes LDA algorithm to the topic keywords extraction technology.At the same time,Chinese semantic is considered in the proposed combination of words generation algorithm.It is mainly to solve the limitations of existing segmentation systems.Finally,relying on entity extraction algorithm,keyword extraction algorithm,sentiment analysis algorithm and multidimensional data display technology,this paper designs and implements special news analysis system.This system can show the highest number of entities and entities related news and articles appear under a particular topic,can show the best performance of a particular topic under the theme of the keyword and a special news of emotion can be a histogram analysis results unfolded form,may be taken in time for clues hot news attention.The research of this paper has the following innovation:in the sentiment analysis algorithm,for news headlines and news text has taken a different algorithm.The title in the sentiment analysis joined the neutral recognition algorithm.The text sentiment analysis joined the subjective recognition and subject word recognition algorithm.