Aspect-Identifcation-and-Extraction-Model

Overview

This repository contains the deliverables for the thesis "Aspect Extraction from E-Commerce Product and Service Reviews," presented to the Department of Software Technology, College of Computer Studies at De La Salle University.

Abstract

Aspect Extraction (AE) is a key task in Aspect-Based Sentiment Analysis (ABSA), but it remains difficult to apply in low-resource and code-switched contexts like Taglish, a mix of Tagalog and English commonly used in Filipino e-commerce reviews. This paper introduces a comprehensive AE pipeline designed for Taglish, combining rule-based, large language model (LLM)-based, and fine-tuning techniques to address both aspect identification and extraction. A hierarchical aspect framework is developed through multi-method topic modeling, along with a dual-mode tagging scheme for explicit and implicit aspects. For aspect identification, a rule-based system demonstrated efficiency and consistency, while the LLM achieved the highest accuracy through few-shot annotation. In contrast, the fine-tuned Gemma-3 1B model showed limited performance, partly due to dataset imbalance and its sensitivity to uneven aspect representation. The LLM was also employed for aspect extraction, demonstrating strong potential in handling Tagalog reviews, particularly for Price and Service aspects, while lower performance in Product and Delivery was linked to Taglish variability, implicit aspect mentions, and context-dependent sentence structures. Overall, the study contributes a scalable and linguistically adaptive framework for enhancing ABSA in diverse, code-switched environments.

---

Proponents & Advisers

Proponents:

• Valiant Lance D. Dionela
• Fatima Kriselle S. Dy
• Robin James M. Hombrebueno
• Aaron Rae M. Nicolas

Adviser:

• Charibeth K. Cheng, PhD
• Affiliation: College of Computer Studies, De La Salle University

Co-Adviser:

• Raphael Gonda, PhD
• Affiliation: College of Computer Studies, De La Salle University

---

Setup Instructions

All setup guides for every technique are included in this repository.
Please refer to the Setup Instructions for All Techniques directory for the complete instructions.

---

Citations

All model details and training setups can be found in our papers. If you use our model or find it useful in your projects, please cite our work:

@article{dionela2025aspectextractionproductservicereviews,
  title={Aspect Extraction from E-Commerce Product and Service Reviews},
  author={Valiant Lance Dionela, Fatima Kriselle Dy, Robin James Hombrebueno, Aaron Rae Nicolas, Charibeth Cheng, and Raphael Gonda},
  journal={ },
  year={2025}
}