Main Article Content
Today we have large amounts of textual data to be processed and the procedure involved in classifying text is called natural language processing. The basic goal is to identify whether the text is positive or negative. This process is also called as opinion mining. In this paper, we consider three different data sets and perform sentiment analysis to find the test accuracy. We have three different cases- 1. If the text contains more positive data than negative data then the overall result leans towards positive. 2. If the text contains more negative data than positive data then the overall result leans towards negative. 3. In the final case the number or positive and negative data is nearly equal then we have a neutral output. For sentiment analysis we have several steps like term extraction, feature selection, sentiment classification etc. In this paper the key point of focus is on sentiment analysis by comparing the machine learning approach and lexicon-based approach and their respective accuracy loss graphs.