Research And Implementation Of Fine-grained Opinion Mining System For Word-of-mouth Monitoring

In recent years,the Internet of Everything,the rapid development of the network,and the continuous popularity of paperlessness have prompted changes in the way people communicate.It is no longer just a face-to-face method.People are more and more keen to express their views and opinions through comments,and the amount of data is increasing.Increment.At the same time,massive data also leads to the continuous accumulation of invalid information,and the problem that effective information is difficult to visualize.In-depth mining of the subjective and effective information contained in these massive comments to extract valuable information can help all industries to grasp public opinion.The situation,to guide the future development of commodities,can also provide users with the necessary key information and make corresponding decisions.Therefore,in order to solve this kind of problem,related systems of word-of-mouth monitoring came into being.Fine-grained opinion mining,as an important part of such systems,plays a very important role in the accuracy of effective information extraction.This article has conducted research and experiments on the fine-grained opinion mining part.It ignores the importance of subjective text extraction for traditional models.When extracting evaluation elements,it does not consider the relationship between words in the sentence and ignores implicit evaluation.The impact of the object on the results of opinion mining and make corresponding improvements.With the help of product review text in the field of skin care products,a new subjective text extraction dictionary was constructed,combined with machine learning methods to construct a model for text classification,and a three-layer evaluation element combination extraction model was proposed,and the results of subjective and objective classification and evaluation element extraction Applied to the realization of sentiment polarity analysis,the sentiment classification of product review text is completed,the classification effects of different models are compared,and the performance of opinion mining model in word-of-mouth detection system is improved.Based on the research results of the opinion mining model,this article applies it to the word-of-mouth monitoring system.The system is based on the Python language and is built using the Flask framework.The Scrapy framework and MongoDB are used to collect and store data.System users can upload existing product review text by customizing to complete the addition of collected data,or configure the collection task through the website link to complete the collection of network data,and then mine the model through opinions,including text preprocessing and subjective text on the collected data Extraction,combination of evaluation elements and emotional polarity analysis,and display the results of opinion mining.