Analyzing Election Sentiments in Tweets with Gated Recurrent Units (GRU)

Abstract

Sentiment analysis, a key task in natural language processing, is important for detecting the emotional tone portrayed in text. In this study, we focus on implementing a Gated Recurrent Unit (GRU) model to analyze attitudes within the 2020 Donald Trump Election tweets dataset. By setting the GRU model with carefully selected parameters, the aim of the study is to unveil the inherent sentiment patterns in the dataset. To develop the sentiment analysis model, the study devised a three phase methodology which that include data preprocessing, feature selection using correlation matrix, and lastly the implementation of GRU. Futhermore, we provided the outcomes of our experiment, evaluating the model's performance through important measures such as accuracy, precision, and recall. Notably, our data exhibit an exceptional accuracy rate of 93%, verifying the model's power to appropriately categorize attitudes. Additionally, both recall and precision receive outstanding ratings of 94% and 96%, indicating the model's skill in distinguishing both positive and negative attitudes. This inquiry emphases the effective usage of the GRU model in sentiment analysis, shedding light on the emotional nuances within the 2020 Donald Trump dataset and enriching our understanding of sentiments during the election period.