[KongchangAI]
ModelQuery Decomposition for RAG

QDRAG

An attributable retrieval-augmented generation framework based on query decomposition. It decomposes complex questions into atomic sub-questions and applies reranking-based pruning to improve citation faithfulness and generalization capability.

Timeline (last 90 days)

Nov 13

A recent paper proposes the QDRAG framework, which systematically addresses RAGQ's over-citation, implicit entity recognition failure, and poor generalization problems through query decomposition.

Unverified50%
Nov 13

The QDRAG framework contains two key steps: query decomposition and reranking-based pruning.

Unverified50%
Nov 13

QDRAG uses a multimodal large language model (MLLM) to decompose the original question into several atomic sub-questions, each corresponding to an independent entity or relation.

Unverified50%
Nov 13

QDRAG proposes two zero-shot query decomposition methods: QD-C (based on contextual knowledge) and QD-R (based on retrieval rewards).

Unverified50%
Nov 13

The paper finds that fine-tuning on the query decomposition task exhibits better generalization than fine-tuning directly on the downstream RAGQ task.

Unverified50%
Nov 13

QDRAG was evaluated on four multimodal question answering benchmarks, covering both in-domain and out-of-domain tests.

Unverified50%
Nov 13

Experimental results show that QDRAG significantly outperforms all baselines on answer grounding and citation faithfulness, even surpassing Gemini-Pro.

Unverified50%
Nov 13

QDRAG experiments show that it effectively reduces over-citation, improves the accuracy of implicit entity recognition, and maintains robust performance in cross-domain scenarios.

Unverified50%

All Facts (8)

Source Articles