A Contextual Framework to Find Similarity Between Users on Twitter

Twitter is one of the most used social networking sites, and people usually prefer to share about themselves, their views, and other things that they have an interest on Twitter. The method proposed can be used by the average Twitter user to find out their degree of similar they are to any other user on the platform. The presented Framework finds similarity between any two users on Twitter dependent on the eight parameters which are Mention Similarity, Common Interest, Topic and List similarity, followers and following relationship similarity, retweets, likes, common hashtags, and profile Similarity. Every parameter generates some score, and the score of each parameter is not dependent on any other parameter score. Weightage has been assigned to each parameter according to the score they are getting individually, and the value of each weight lies between 0 and 1. Each parameter requires user data that has been extracted using Twitter’s own API such as follower, retweet, like, hashtag, etc. For each Twitter user, data of eight parameters are collected for 2019 October to 2020 October. The framework can be used for suggesting how similar two users on Twitter are. The framework has been verified using datasets of five users, and from these datasets, percentage similarity is being calculated. For finding the effectiveness of the framework, the result of our case study was compared against a survey of human judges consisting of 524 people and was found to be moderately effective.