Multi Trend Twitter Sentiment Analysis: Collaborative Approach for Improved Results

Abstract

Twitter has a significant number of daily users through which tweets are utilized to communicate their thoughts in this era of growing social media users. This paper offers a way to haul out sentiments from tweets as well as a method for sorting out various tweets as optimistic, adverse, or unbiased. It refers to identifying and classifying the sentiments expressed in the text source. The existing Twitter APIs for data extraction are used to mine public Twitter data. Tweets would be chosen based on a few carefully chosen keywords related to the domain of our concern. In our proposed method, we collected various sentiment data from a variety of tweets to train and produce more precise and reliable sentiment classifiers for each trend. This method automatically extracts the key elements of subjects from online user evaluations. Since tweets are generally unstructured in format, they must first be converted into structured format. And after that, the data is fed into several models for training and used to rank the best sentiment classifier. The intention of this design is to arrive at a model that can classify sentiments of real-world data using Twitter.