Learning to Extract Cross-Domain Aspects and Understanding Sentiments Using Large Language Models
Authors:
Karukriti Kaushik Ghosh,
Chiranjib Sur
Abstract:
Aspect-based sentiment analysis (ASBA) is a refined approach to sentiment analysis that aims to extract and classify sentiments based on specific aspects or features of a product, service, or entity. Unlike traditional sentiment analysis, which assigns a general sentiment score to entire reviews or texts, ABSA focuses on breaking down the text into individual components or aspects (e.g., quality,…
▽ More
Aspect-based sentiment analysis (ASBA) is a refined approach to sentiment analysis that aims to extract and classify sentiments based on specific aspects or features of a product, service, or entity. Unlike traditional sentiment analysis, which assigns a general sentiment score to entire reviews or texts, ABSA focuses on breaking down the text into individual components or aspects (e.g., quality, price, service) and evaluating the sentiment towards each. This allows for a more granular level of understanding of customer opinions, enabling businesses to pinpoint specific areas of strength and improvement. The process involves several key steps, including aspect extraction, sentiment classification, and aspect-level sentiment aggregation for a review paragraph or any other form that the users have provided. ABSA has significant applications in areas such as product reviews, social media monitoring, customer feedback analysis, and market research. By leveraging techniques from natural language processing (NLP) and machine learning, ABSA facilitates the extraction of valuable insights, enabling companies to make data-driven decisions that enhance customer satisfaction and optimize offerings. As ABSA evolves, it holds the potential to greatly improve personalized customer experiences by providing a deeper understanding of sentiment across various product aspects. In this work, we have analyzed the strength of LLMs for a complete cross-domain aspect-based sentiment analysis with the aim of defining the framework for certain products and using it for other similar situations. We argue that it is possible to that at an effectiveness of 92\% accuracy for the Aspect Based Sentiment Analysis dataset of SemEval-2015 Task 12.
△ Less
Submitted 15 January, 2025;
originally announced January 2025.
Atom Search Optimization with Simulated Annealing -- a Hybrid Metaheuristic Approach for Feature Selection
Authors:
Kushal Kanti Ghosh,
Ritam Guha,
Soulib Ghosh,
Suman Kumar Bera,
Ram Sarkar
Abstract:
'Hybrid meta-heuristics' is one of the most interesting recent trends in the field of optimization and feature selection (FS). In this paper, we have proposed a binary variant of Atom Search Optimization (ASO) and its hybrid with Simulated Annealing called ASO-SA techniques for FS. In order to map the real values used by ASO to the binary domain of FS, we have used two different transfer functions…
▽ More
'Hybrid meta-heuristics' is one of the most interesting recent trends in the field of optimization and feature selection (FS). In this paper, we have proposed a binary variant of Atom Search Optimization (ASO) and its hybrid with Simulated Annealing called ASO-SA techniques for FS. In order to map the real values used by ASO to the binary domain of FS, we have used two different transfer functions: S-shaped and V-shaped. We have hybridized this technique with a local search technique called, SA We have applied the proposed feature selection methods on 25 datasets from 4 different categories: UCI, Handwritten digit recognition, Text, non-text separation, and Facial emotion recognition. We have used 3 different classifiers (K-Nearest Neighbor, Multi-Layer Perceptron and Random Forest) for evaluating the strength of the selected featured by the binary ASO, ASO-SA and compared the results with some recent wrapper-based algorithms. The experimental results confirm the superiority of the proposed method both in terms of classification accuracy and number of selected features.
△ Less
Submitted 10 May, 2020;
originally announced May 2020.
