Knowledge Graph Prompting: A New Approach for Multi-Document Question Answering
Multi-document question answering (MD-QA) involves answering questions that require synthesizing information across multiple documents.
This is an important capability for real-world information retrieval systems.
However, most existing question answering techniques are designed for either open-domain QA with a large corpus or single document QA.
In a new paper, researchers from Vanderbilt University and Adobe Research propose a novel approach called Knowledge Graph Prompting (KGP) to address the challenges of MD-QA.
Challenges in Multi-Document QA
MD-QA introduces some unique challenges compared to other QA tasks:
- Multi-hop reasoning: Answering a question may require logical reasoning across multiple passages from different documents. For example, bridging entities across documents.
- Retrieval latency: Retrieving useful passages from a large collection of documents is time-consuming.
- Diverse modalities: Documents may contain varied structures like paragraphs, lists, tables, images etc. Reasoning across these modalities…