Hybrid Sentiment Classification of Reviews Using Synonym Lexicon and Word embedding

Sentiment analysis is used in extract some useful information from the given set of documents by using Natural Language Processing (NLP) techniques. These techniques have wide scope in various fields which are dealing with huge amount of data link e-commerce, business and market analysis, social media and review impact of products and movies. Sentiment analysis can be applied over these data for finding the polarity of the data like positive, neutral or negative automatically or many complex sentiments like happiness, sad, anger, joy, etc. for a particular product and services based on user reviews. Sentiment analysis not only able to find the polarity of the reviews. Sentiment analysis utilizes machine learning algorithms with vectorization techniques based on textual documents to train the classifier models. These models are later used to perform sentiment analysis on the given dataset of particular domain on which the classifier model is trained. Vectorization is done for text document by using word embedding based and hybrid vectorization. The proposed methodology focus on fast and accurate sentiment prediction with higher confidence value over the dataset in both Tamil and English