What to Predict from Twitter Data?

Abstract

In the last decade, Twitter data has become one of the most valuable research sources for many areas including health, marketing, security, and politics. Researchers prefer Twitter data since it is completely public and can be easily downloaded using Twitter APIs. The recent intensive use of Twitter data makes it difficult for researchers to follow or analyze its research. In this paper, we summarize most of the predictable patterns, aspects, and attitudes from Twitter data and analyze the performance and feasibility of the algorithms used. Moreover, we describe the current popular Twitter datasets used in various domains and applications. Current challenges and research gaps are discussed, and some recommendations are given for future works from different perspectives.