A Categorical Approach on Sentimental Analysis | Original Article

ABSTRACT:

In today’s age of internet, web 2.0 technology is very prevailing. Now people can easily express their opinion on various topics like reviews on a newly launched product or movie through platform such as blog, spaces, forums and twitter etc. More than 20 billion pages have textual data over the internet. For this reason companies need to filer this text and calculate the cognizance for business. The classification of Sentiment is a technique to focus on the opinions on a particular article by people or conveyed orally. The term sentiment includes emotions, conclusions, behavior and others. In this thesis, main focus is on human readable text writing on the e-commerce sites. Sentiments or opinion of people can be majority of two types positive and negative. First, in order to check the polarity of emotions, various steps were taken. The way to deal with emotions is to clean up the data, including removing stop words, spaces, repeated words, emojis and hash tags. In order to correctly classify these views, machine learning techniques are used. There are several methods that can be used to extract features from the source text. It is divided into 2 stages In the first stage, the extraction of data related to the opinion is completed, that is, the data specific to the opinion is extracted. Now by doing this, the opinion is converted to normal text. In the next stage, more features will be extracted and added to the feature vector. Each opinion in the training data is associated with a specific class label. Pass the training data to different classifiers, and then train the classifiers. After the test is completed, the opinions are provided to the model and the classification is completed with the help of these well-trained classifiers. Therefore, in the end, opinions divided into n different categories can be obtained.