A CONCEPTUAL AQUILA MERGED ARITHMETIC OPTIMIZATION (AIAO) INTEGRATED AUTO-ENCODER BASED LONG SHORT TERM MEMORY (AUE-LSTM) FOR SENTIMENT ANALYSIS

Abstract

Sentiment analysis is a branch of analysis that uses disorganized written language to infer the opinions and emotions of people's critiques and attitudes toward entities and its features. In order to produce acceptable results, the majority of sentiment analysis models that employ supervised learning algorithms require a large amount of labeled information during the training stage. This is typically costly and results in significant labor expenses when used in practical applications. In this study, an intelligent and unique sentiment prediction system is developed for accurately classifying the positive, negative, and neutral comments from the social media dataset. Data preprocessing, which entails noise reduction, tokenization, standardization, normalization, stop word removal, and stemming, is done to ensure that the data is of a high enough quality for efficient sentiment prediction and analysis. The preprocessed data is then used to extract a mix of features, including hash tagging, Bag of Words (BoW), and Parts of Speech (PoS). Consequently, in order to choose the best features and speed up the classifier, a new hybrid optimization method called Aquila merged Arithmetic Optimization (AIAO) is used. Furthermore, an Auto-Encoder based Long Short Term Memory (AuE-LSTM), an innovative and clever ensemble learning technique, is used to precisely anticipate and classify user feelings based on the chosen data. This study uses a variety of open source social media datasets to evaluate the performance of the suggested AIAO integrated AuE-LSTM model.