Research On The Electronic Commerce Comment Analysis System Based On Data Acquisition

With the continuous development of Internet technology and the steady development of logistics industry,online shopping has become the preferred shopping mode for more and more people.The evaluation information of products reflects users' opinions and attitudes towards products and has high value.On the one hand,the comment information of the product can help other users to provide a certain purchase guidance,on the other hand,the targeted evaluation of a certain aspect of the product can effectively improve the product and improve the qualit y of service.However,it is very difficult to dig out the required information from a large number of product evaluation contents.In view of the above problems,this paper studies the collection and excavation of the electricity supplier evaluation.First,we collect the evaluation data of the e-commerce website and store it quickly.Based on the Scrapy framework,this article takes the evaluation information of Jingdong mall as a crawling object.In order to meet the needs of the evaluation system,the crawling strategy is improved.In view of the access restrictions and Robot protocols,cookie and user-agent are used to circumvent restrictions.To meet the fast read and write of database,Mongo DB is used for storage.Before evaluating information mining,Chinese word segmentation is first processed.In this paper,based on BI-LSTM,the LSTMN unit is used to replace neurons,and BI-LSTMN-CRF model is proposed combined with CRF model.Dropout is used to prevent overfitting during training.The result shows that the accuracy is improved.Secondly,we use the LDA model to extract the subject and inclusion of the processed text,and enrich the local corpus with the network resources,then use the three layer CRF model to set the corresponding feature rules,and classify the emotional attitude and emotional intensity of the text respectively.Through accuracy analysis,this method can effectively deal with the text.Finally,a comment analysis system is designed.Based on the JFinal framework,the above algorithms are integrated and applied to the system.Under the product scoring rules,the products are sorted and displayed after been graded,and the graphical display of users' concerns is carried out.After testing,the algorithm used in this paper we can achieve the goal of product analysis,and we can objectively display product characteristics and user needs.