Sentiment Analysis Using Supervised Machine Learning Technique
Jitendra Kumar · 2017
At present time, People are free to share their thoughts and idea over the Internet. These thoughts and comments help to understand the current scenario of a person, country or mass. But the problem is that the available data is huge to understand. So to summarize the data and to get a proper essence of the data an approach is needed that is called Sentiment Analysis. It is all about to analyze the human comments, opinions, ideas or sentiments. Sentiment analysis is having various application in different domain due to the exponential growth of Internet users. Different approaches are their to analyze the human comments or sentiments. In first work, CountVectorization (CV) and Term Frequency-Inverse Document Frequency (TF-IDF) approach have been used to assign the numerical value to each word. After that different Machine Learning Classifiers are used to classify the sentiments. Support vector machine (SVM) gives good result and includes less error. In second work, SentiWordNet is used to assign the weight to each word. The contributions of the work are: (a) Generate feature weight model using SentiWordNet 3.0, (b) assign feature weight to every feature according to POS tag, and (c) suitable combination selection of adj, adv, verb and noun to give better results. The features are applied to the classifier and 74% accuracy is achieved using SVM with 10 folds. Four Performance metrics are used accuracy, precision, recall, and F1-score.