A Lexicon-based Method to Determine Subjectivity of Unstructured Data

Twitter has become a popular micro-blogging platform for people from academicians to politicians, business-oriented people to celebrities. Business companies perform analysis for public opinions about their products and services. Likewise, politicians do it for understanding their popularity among people for their policies. Twitter provides platform to online users for sharing their opinions about certain events, news, and trends around the world. Thus, leading to generation of massive unstructured data (i.e., textual data). This paper presents determination of subjectivity (i.e., opinions) embedded in unstructured data by means of lexicon-based method. The study considered three different datasets: 1) Dataset₁ â crawled from Twitter based on certain topics, 2) Dataset₂ â user reviews about three websites (imdb.com, amazon.com, yelp.com) and 3) Dataset₃ - review comments about research articles reviewed by experts. The Dataset₂ and Dataset₃ also contained expert opinions; therefore, subjectivity of these datasets has been compared with expert opinions to validate the applied method. The outcomes showed significant potential and effectiveness of the lexicon-based method.