Publication

Abstract : Social media is an online platform where users freely share their views on various topics. By sharing their views, they come to a mutual agreement or disagreement with other users’ views. Users are quite active on social media and the number of users is growing exponentially with time. This leads to an abundance of data available through social media. This data helps us in curating different research problems. One such research problem can be finding similarities between users by analyzing user data. Data on social media is available in many forms like textual posts, comments, likes, shares, etc. Users often tend to express their diversity of opinions widely through comments on social media. Many opinionated and active users on social media are found in the comments section under a post. Hence, comments play a major role in determining similarity in user judgment about a topic. Likewise, people liking a post is also an essential feature in finding similarities between users. People with similar opinions tend to like posts of a specific type. This paper talks about how comments, textual posts, and likes for comments can be combined to form something called a keyword or a tag. These tags that are extracted from user data are further used in building a tag network. The tag network is used to make communities of users with similar interests. User grouping can also be done based on these extracted tags and comments under the posts. The proposed methods in this paper to achieve the above-mentioned are TF-IDF (Term frequency-inverse document frequency) followed by the TextRank algorithm.