Automated Sentiment Analysis Using Pigeon Inspired Optimization Algorithm with Attention Bidirectional Gated Recurrent on Social Media

Sentiment Analysis (SA) is a model employed in Natural Language Processing (NLP), which describes the feeling or sentiment conveyed in textual content. It uses automated tools to identify psychological data like thoughts, attitudes, and mental states shown in writing and indirectly over social network platforms. SA model is regularly executed on text datasets to help understand user desires, responses, and businesses observing products. Therefore, the Deep Learning (DL) technique has been developed as a promising model that has been commonly used and reached significant results. As DL approaches mechanically eliminate features from the database, there exists a possibility that an intermediate symbol, which has been disregarded may serve as a related factor. This study introduces an automated SA using the Pigeon-Inspired Optimization Algorithm with Attention Bidirectional Gated Recurrent Unit (PIOA-ABGRU) technique for social networking. The presented PIOA-ABGRU technique mainly focused on detecting multiple classes of feelings that occur in social media. The PIOA-ABGRU approach undergoes various sub-processes to alter the input data into a beneficial layout. Also, the Word Embedding (WE) procedure is performed using the BERT approach. For sentiment detection, the PIOA-ABGRU technique applies the ABGRU model, which detects sentiments in distinct classes. Finally, the PIOA-based hyperparameter tuning procedure is executed to select ABGRU's hyperparameters. The simulation results of the PIOA-ABGRU model take place on a social media dataset. The experimental analysis specified that the PIOA-ABGRU model reaches effectual achievement over other models using distinct measures.