Computer Big Data Sentiment Intelligent Analysis Using Maximum Likelihood and Deep Learning

This With the field of machine learning continues flourishes, the rising demand of understanding human emotions brings attention of solving the issue of SA using deep learning. In this study, we want to utilize CNN techniques to evaluate a real dataset containing movie reviews with annotations either positive or negative. We will clean the data by removing html tags, removing non-alphabetical word, and removing stop words. Then we will first introduce one classification method based on statistical analysis based on maximum likelihood and discuss the usage of different smoothing methods. We will be able to construct a bag-of words model based on the statistical analysis. Lastly, we will convert the vector into multi-dimensional matrix and build a CNN model to categorize whether a movie review is positive or negative. As a result, we show that both approaches are effective proved by cross validation and discuss the differences between two methods and some common drawbacks.