FARSum: Feedback-Aware Abstractive Product Review Summarization

Retail customers read through multitude of online product reviews to make confident purchase decisions. To automate this process, we explore and evaluate several state-of-the-art (SOTA) models for summarizing product reviews along three dimensions: a summary product verdict, pros, and cons. To improve the performance of summarization from a large number of reviews per product, we propose FARSum, an efficient solution that leverages review filtering based on review recency and customer feedback including review helpful vote and review rating in the first stage. To improve context generalization across product categories, we train a BART-based model on synthetic review summaries and fine tune the model using ground-truth summary labels. We demonstrate the competitive performance of our solution vis-a-vis other SOTA models using the ROUGE metrics.