Evaluation of Deep Learning Approaches for Aspect Based Sentiment Analysis on Movie Dataset

Aspect-Based Sentiment Analysis (ABSA) permits users to compute the sentiment for an aspect in a specific context. The domain-specific knowledge plays a major role in analyzing the sentiments of aspects. This res earch work dis covers the probability of enhancing the knowledge-driven ABSA concerning effectiveness and efficiency. These problems can be tackled here by building the comparative analysis on different machine learning algorithms on movie data. The main intent of this work is, to analyze the sentiment analysis on movie datasets to conduct comparative analysis through classifiers like SVM, DT, NB, KNN, and NN, where the classification accuracy can be improved by extracting the s ignificant features and aspects from the contexts. In the input movie data, s top words removal, removing punctuation, lower case conversion, and stemming are performed as the pre-processing step. The opinion words are collected in the as pect extraction phas e, and further, the features are extracted via determining the polarity s core and word2vector. The extracted features are considered as the input to the different machine learning algorithms like K-Neares t Neighbor (KNN), Decis ion Tree (DT), Naive Bayes (NB), Support Vector Machine (SVM), and Neural Network (NN). Experiment res ults and in-depth analys ess how that the five approaches yield better attention towards enhancing the performance.