Design And Implementation Of Comment Sentiment Analysis System Based On Machine Learning

With the rapid popularization of mobile Internet,the purpose of people using the network has changed greatly.The emergence of e-commerce platform has also greatly changed people's consumption mode,and Internet comment information also gave people more reference before they make decisions.In terms of the use process of hotel reservation platform,users can get more intuitive judgment by browsing the hotel's comment information.Through the sentiment analysis of the hotel comment data,we can clearly understand the user's objective evaluation of the hotel and the check-in experience,so as to help other users make quick decisions,choose the right hotel to stay,and reduce the user's choice time cost.For the task of Chinese text sentiment analysis,the quality of the results,to a great extent,depended on the construction of sentiment features.Due to the diversity of Chinese text expression forms,sentiment features often have some problems,such as high spatial dimension,sparse representation,lack of text semantic information and so on.In order to solve the above problems,this dissertation combined the machine learning algorithm and the language model of natural language processing,studied the algorithm of text sentiment feature extraction and expression,and uses machine learning model to carry out sentiment polarity.And based on the proposed sentiment analysis model,this dissertation designed and implements a hotel comment sentiment analysis system.Users can query the hotel distribution map,hotel satisfaction map,and the sentiment distribution of hotel comments in Taiyuan through this system.Most importantly,this system can conduct real-time sentiment analysis for user's hotel comment information.Specifically,this dissertation mainly includes the following three aspects of work:(1)Construction of Hotel Comment Corpus: due to the deficiency of standard datasets in the process of actual sentiment analysis task,this dissertation combined the existing web crawler principle and related technical methods to collect and store the hotel comment information of hotel.qunar.com,and constructed a large-scale hotel comment corpus of positive and negative sentiment categories.(2)Research on feature extraction and representation of Chinese text: this dissertation combined the differences of different word vectors in text representation,and combined the advantages of multiple word vectors.This dissertation clearly introduced the implementation process of RS-Word2vec?Glove(RS-Wv Gv),a feature representation method of sentiment words based on rough set and multi-channel word vectors.(3)Design and implementation of hotel comment sentiment analysis system: this dissertation designed and implemented a hotel user comment sentiment analysis system based on machine learning,and applied the proposed Chinese text sentiment analysis model into the relevant modules of sentiment analysis system,the system can meet the needs of user's data visualization and sentiment analysis.