Research On Sentiment Classification Model Based On Web Comments

With the highly developed network today,more and more social networks and e-commerce platforms have penetrated into every aspect of life.People are not only recipients of network information,but also producers of information.The review data generated by many netizens can dig out huge social and commercial values.By analyzing the sentiment tendency of these data,government departments can learn the public opinion trend of a major event in a timely manner,users can understand the characteristics of the product,merchants can also learn the needs of users,understand the shortage of products,and adjust marketing strategies in time.Existing classification methods mainly include sentiment dictionary methods and machine learning methods.The sentiment dictionary method relies too much on the sentiment words in the dictionary.The more complete the sentiment dictionary,the more pronounced the sentiment tendency of online comments and the better classification effect.The classification effect of comments is not good when the sentiment tendencies are not easy to distinguish.The machine learning method is a supervised method,and its classification effect relies on a large number of pre-annotated corpora.Currently,the corpus annotation is done manually,and the workload is extremely large.This paper combined characteristics of the two methods to build a new sentiment classification model of network reviews.First,expand the sentiment dictionary based on the domain of online reviews,and calculate the sentiment value of each online comment according to the extended sentiment dictionary.According to the preset sentiment threshold,the comments with significant sentiment tendencies and higher accuracy were selected as the definite set,and the rest that were not easily distinguished as uncertain sets.The classification result of the definite set was directly determined by the sentiment value.Second,according to the definite set from the sentiment dictionary method,a classifier was trained through self-supervised learning,and the training data did not require manual annotation.Finally,the trained classifier was used to classify the uncertain set again,and an improved algorithm was used to improve the classification result of the uncertain set.The main research contents are as follows:(1)Research on sentiment classification based on sentiment dictionary method.Based on the existing sentiment dictionary,the field dictionary is expanded.The sentiment value is calculated based on the dictionary and rules,and then classified according to the sentiment value.The experiment found that the more complete the dictionary,the higher the absolute value of sentiment value,the higher the accuracy rate,and the effect of sentiment classification on negative reviews is better.(2)Research on sentiment classification based on machine learning methods.Use word2 vec for vector expression of network comments,and then use PCA algorithm to reduce dimensionality,as the input of machine learning classification algorithm.Experiments show that the support vector machine classification algorithm performs best in this method,and the judgment correct rate of positive comments is higher than that of negative comments.(3)A classification model that combines sentiment dictionaries and machine learning.The sentiment dictionary is used to calculate the sentiment score of each comment.Based on the sentiment value,the sentiment is clearly determined and the sentiment is ambiguous.The determined set is used as the machine learning training corpus and the uncertain set is used as the test corpus.The classification result shall prevail.The classification result of the uncertain set is corrected by combining the two methods.The experimental verification shows that the accuracy of the model is the highest in the three data sets.