Research On Sentiment Analysis Algorithm Based On Word Vector And POS

Sentiment analysis is used to study events and people's views on events.As a communication tool,micro-blog has attracted great attention of experts and scholars at home and abroad.It is very important and meaningful to study the emotional analysis of micro-blog data.The traditional sentiment analysis usually uses sentiment dictionary to analyze the sentiment tendency of text.On the one hand,traditional sentiment analysis ignores that the non-sentiment words may affect the results of sentiment analysis;on the other hand,the traditional semantic model can not store the word's semantic information,so it will result that sentiment analysis cannot be accurately classified.In order to solve the above problems,we propose the Sentiment Analysis Algorithm based on Word Vector and POS(SA2-WV&POS).The main work of this paper is as follows:(1)In the preprocessing stage,in view of the possible problems arising from sentiment analysis using sentiment dictionary,this paper processes the dataset by part of speech(POS)filtering.On the basis of taking full account of affective words,this method takes into account the influence of non-sentiment words on sentiment analysis.(2)In the feature extraction phase,the traditional semantic model can't store the word's semantic information.In this paper,we use Word2 Vec to map words to higher dimensional and sparse vector spaces.The SA2-WV&POS Algorithm uses the Word2 Vec to convert word into word vector.This algorithm combines the TF-IDF feature extraction method with the word vector,it not only takes full account of the semantic information of words,but also can control the dimension of word vector.The SA2-WV&POS Algorithm fully considers the effect of POS and semantic information.The algorithm combines the TF-IDF with the word vector to generate the feature word vector.The experiment shows that the algorithm obviously improves the accuracy and F-measure etc,and gets better results of sentiment analysis.