Automatic Classification And Extraction Of Expert Opinions Based On Network

With the development of science and technology and the improvement of living standard,more and more goods are coming into our life.Products such as automobiles and electronic products are characterized by high price and high technology.Since higher prices lead to higher opportunity costs for repeated purchases,consumers are more cautious in their purchases.But the product itself high technical content,the consumer is very difficult to rely on their own ability to judge the relative merits,so the relevant experts in the field of view has become an important channel for the consumers to obtain useful information.Through the Internet anyone can quickly and cheap access to a large number of articles,but these articles different style,different views,some views are even factors related to the interests of the left and right sides.The old saying is: listen to both sides.But because of the huge amount of information,it is difficult for non-professionals to extract information that is useful to them.Car review article,for example,this study by analyzing articles classification standard,and all kinds of articles feature extracting,unearth hidden behind all kinds of articles,finally collected all kinds of ideas in an objective and comprehensive conclusion,the purpose is to provide reliable basis for consumers to purchase goods.In the concrete research process,first through the web crawler technology implements data acquisition,secondly according to the LDA model and word frequency statistics analysis the parameters of the motor,finally select engine,configuration,space,in this paper,the main research aspects and so on six big variables.Then,the paper quantifies the text data using the emotion analysis method,and constructs the emotional entity word table applicable to the automobile field in the process of quantification.After obtaining the numerical data,by comparing the results of various clustering methods,the classification of articles by kmeans clustering method was finally determined.In the end,the characteristics of various articles are extracted and collected,and the scoring formula is determined to get the final score of the car.The significance of this study is to review article by car,for example,by studying the formation of a set of score calculation process,and to help consumers to simplify the process of access to information on the Internet,enable consumers to less energy input to obtain more accurate,more helpful to the decision-making of valid conclusions,thus promoting a set of feasible and can be applied to stock,electronic products and other text data processing in the field of process.