A Lightweight Aspect-Based Sentiment Analysis for Food Delivery Reviews Using Dependency Parsing

Abstract: As digital platforms continue to shape consumer behaviour, the suitability of ordering food online has led to a surge in user-generated reviews. Interpreting user perspectives has become essential, and sentiment analysis offers a structured method to uncover emotions conveyed through textual data. A more granular technique, Aspect-Based Sentiment Analysis (ABSA), identifies sentiment specific to particular aspects of a service or product. This study introduces a lightweight, rule-based ABSA framework tailored to food delivery reviews. The model incorporates methods like dependency parsing and negation recognition to enhance sentiment detection. From a collection of 2,345 reviews, a subset of 40 was manually annotated to assess performance. The system achieved strong metrics in terms of precision, recall, and F1-score, representing its fitness for use in domains with limited annotated data or computational resources.

Keywords: Aspect-Based Sentiment Analysis, Dependency Parsing, Opinion Mining, Sentiment Polarity