PROCREATION AND APPLICATION OF SENTIMENT BASED DICTIONARY TO REVEAL POPULAR EVENT TOPICS TRENDS ON TWITTER PLATFORM

Abstract

The layers of information are formed by the rapid rise of the data storm that is generated on Social Media(SM). The work of finding hidden trends is tough owing to the constant flow of tweets caused by volume and variety. The study recommends the use of dictionaries to absorb information more quickly and filter out inadequate data. The research proposes the procedural method to build the global dictionary. The global dictionary is created using the datasets on social, political, scientific and sports domains on most recent events to cover major sentiment drifts. This dictionary can help in revealing trends which can further be used to detect the event trends. The suggested model performs trend analysis on a tweet-by-tweet basis using dictionaries. A tweet-specific trend evolves into a day trend for an event. The proposed model aids in the examination of trends for an event on a timeline. The event trend has the ability to reveal the event's future trajectory. The model is validated using a multi-event dataset from Event 2012 and evaluated successfully on a Citizenship Amendment Act (CAA) dataset from 2019. The model performs well across a wide range of datasets.