Analysis and evaluation of twitter data for stance detection and mobile app recommendation by topic modeling using clustering techniques

Abstract

Micro-blogging has now emerged as a popular conversation and information search tool. Twitter is one of the most famous micro-blogging platforms with over five hundred million active users. Twitter contains numerous information such as a tweet, retweet, account users, account bio, Geo coordinates, user language, source, profile image URL, user followers count, user-friend count, user location, status URL and entity string. This information is associated with the tweets messages and users. Each tweet has a creator, a message, a completely unique id, a timestamp of when it turned into posted, and occasionally Geo metadata, shared by the user. Each user has a Twitter name, an ID, a number of followers, and very often an account bio. Tweets are the basic atomic building blocks of all things on Twitter. Tweets are also called status updates . The tweet factor has a protracted listing of root-stage attributes, which includes essential attributes such as identity, timestamp, and text. Here, identity has a representation of the unique identifier for this tweet and timestamp has UTC time, which tells when this tweet was created and text has an actual UTF-8 text of the status update. The user contains public Twitter account metadata and describes the tweet. Users can be anyone or anything. They tweet, retweet, add quotes to tweet, follow others, create lists, have a home timeline, may be referred to and can appear in bulk. These user metadata values are exceptionally constant. Tweet text has short and less than 140 characters (now 280 characters), assorted with inferred traces. Twitter bio has a length of fewer than 160 characters and it is an accurate one that tells what the user really does, in which how one can pitch the true identity of the Twitter users. Twitter offers a bio-box in which users can provide data on themselves in less than 160 characters. newline