RAG-based Question Answering over Heterogeneous Data and Text

TL;DR

The paper introduces QUASAR, a unified RAG-based question answering system that effectively handles heterogeneous data sources with high accuracy and lower computational costs.

Contribution

It presents a novel unified architecture for QA over diverse data types, incorporating question understanding and evidence filtering to improve answer quality.

Findings

Achieves comparable or better accuracy than large GPT models.

Reduces computational cost and energy consumption significantly.

Demonstrates effectiveness across multiple benchmark datasets.

Abstract

This article presents the QUASAR system for question answering over unstructured text, structured tables, and knowledge graphs, with unified treatment of all sources. The system adopts a RAG-based architecture, with a pipeline of evidence retrieval followed by answer generation, with the latter powered by a moderate-sized language model. Additionally and uniquely, QUASAR has components for question understanding, to derive crisper input for evidence retrieval, and for re-ranking and filtering the retrieved evidence before feeding the most informative pieces into the answer generation. Experiments with three different benchmarks demonstrate the high answering quality of our approach, being on par with or better than large GPT models, while keeping the computational cost and energy consumption orders of magnitude lower.