An Approach to Track Context Switches in Sentiment Analysis

Ever-increasing social media platforms provide colossal amounts of opinionated data. Sentiment analysis on social media is a valuable tool for the process of understanding these new means of online expression, detecting the relevant ones, and analyzing and exploiting them appropriately. Through this work, we introduce an innovative approach for separating text that conveys more than one theme. It focuses on efficiently segregating these different themes, sometimes known as context switches, and then accurately mining the different opinions that may be present in the text containing context switches. We utilize three categories of features namely positional, lexical semantic, and polarity features for theme-based text segmentation within a document. Themes of all the segments are obtained by using a simple noun phrase extractor and sentiment analysis on the different segments is performed to extract the opinions. We also propose an application that improves the efficiency of sentiment analysis and illustrates its working on two sample opinionated documents.