Research On Comment Mining System Based On Sentiment Analysis

In recent years,with the development and popularization of network technology,the amount of data on the network has increased geometrically,and it is unrealistic to collect consumers’ emotional attitudes towards commodities by artificial means.Using the product comment mining system to intelligently obtain comment information about goods and services,and analyze the emotional tendency of these comments,consumers can use them as shopping reference,while merchants can use them to improve goods and services and gain competitive advantage.In this dissertation,the process and key technologies of sentiment analysis are studied,and the text feature extraction method is emphatically analyzed.Aiming at the problems of large computation,low extraction efficiency and insufficient representativeness of the extracted feature words,a feature extraction method based on word frequency and semantics is proposed.The feature word set is expanded by similarity calculation,and the final commodity feature dictionary is obtained after manual classification.In the process of data processing,multi-threaded crawler is used to obtain data,evaluation function is designed according to the characteristics of spam comments,spam comments are filtered by Naive Bayesian classifier,Chinese word segmenter is investigated and studied,and the word segmenter suitable for this system is selected.In the classifier module,three commonly used classifiers,Naive Bayes algorithm,Support Vector Machine(SVM)algorithm and k-nearest neighbor algorithm,are used to train commodity features.Through experimental comparison,finally,Support Vector Machine(SVM)is selected as the classifier of this system.According to the above algorithm,a comment mining system based on sentiment analysis is designed and implemented,and the overall structure and functional modules of the system are designed and implemented.The system extracts,cleans,extracts features and classifies emotional tendencies of comments from online shopping platforms.Experimental tests show that the system can achieve the purpose of sentiment analysis of commodity reviews,and the analysis shows that the feature extraction method proposed in this dissertation has achieved good experimental results.