A Data-Driven Consumer Insight Model Using Aspect-Based Sentiment Analysis and Probabilistic Classification via Naïve Bayes
Year: 2026
Type: article
Abstract: The rapid expansion of e-commerce platforms has significantly transformed consumer purchasing behavior, making online product reviews a crucial source of information for decision-making. However, the large volume and unstructured nature of these reviews pose challenges for both consumers and sellers in effectively evaluating product quality. This study proposes an intelligent online product evaluation system that enhances customer decision-making by integrating Aspect-Based Sentiment Analysis (ABSA) and the Naive Bayes classification algorithm. The system automatically collects product reviews from major e-commerce platforms such as Shopee, Amazon, and Lazada and processes a dataset of 6,736 labeled reviews. Text preprocessing techniques, including tokenization, stop-word removal, and lemmatization, were applied, followed by feature extraction using Term Frequency–Inverse Document Frequency (TF-IDF). The Naive Bayes algorithm was utilized to classify sentiments as positive or negative, while ABSA identified sentiments related to specific product aspects. To improve interpretability, the system incorporates abstractive summarization and data visualization techniques to present concise insights. Experimental results show that the sentiment classification model achieved approximately 80% accuracy, demonstrating the effectiveness of Naive Bayes. System quality evaluation using the FURPS model yielded an overall mean score of 4.24, indicating high user acceptance. The findings confirm that combining ABSA with Naive Bayes provides an efficient, interpretable, and reliable approach to automated product evaluation, offering valuable insights for both consumers and online sellers.
Institution: FEU Alabang
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7224840478