Search papers, labs, and topics across Lattice.
This paper introduces a hybrid approach for aspect-based sentiment analysis that leverages LLMs for aspect identification and traditional machine learning for sentiment classification to address scalability challenges in analyzing large volumes of customer reviews. The authors used ChatGPT to identify key aspects from a sample of restaurant reviews and then trained sentiment classifiers on human-labeled data, applying these classifiers to 4.7 million reviews. Their regression analysis demonstrates that machine-labeled aspects significantly explain variance in overall restaurant ratings across various dining experience factors, cuisines, and regions.
LLMs can bootstrap scalable aspect-based sentiment analysis by identifying key aspects, which can then be used to train efficient classifiers on massive datasets.
Customer-provided reviews have become an important source of information for business owners and other customers alike. However, effectively analyzing millions of unstructured reviews remains challenging. While large language models (LLMs) show promise for natural language understanding, their application to large-scale review analysis has been limited by computational costs and scalability concerns. This study proposes a hybrid approach that uses LLMs for aspect identification while employing classic machine-learning methods for sentiment classification at scale. Using ChatGPT to analyze sampled restaurant reviews, we identified key aspects of dining experiences and developed sentiment classifiers using human-labeled reviews, which we subsequently applied to 4.7 million reviews collected over 17 years from a major online platform. Regression analysis reveals that our machine-labeled aspects significantly explain variance in overall restaurant ratings across different aspects of dining experiences, cuisines, and geographical regions. Our findings demonstrate that combining LLMs with traditional machine learning approaches can effectively automate aspect-based sentiment analysis of large-scale customer feedback, suggesting a practical framework for both researchers and practitioners in the hospitality industry and potentially, other service sectors.
