🤖 AI Summary
Existing RAG paradigms neglect the cognitive application of knowledge in task-specific reasoning, leading to a misalignment between retrieved facts and actual inference requirements. To address this, we propose an application-aware RAG framework that explicitly embeds task-oriented reasoning into the retrieval-generation pipeline. Our approach introduces a bilingual corpus—comprising a knowledge base and task-aligned application examples—supporting both manual and automated construction. We design a modular architecture featuring dual-path joint retrieval, example-alignment modeling, and multi-domain prompt-driven LLM collaborative reasoning. This is the first work to achieve end-to-end integration from retrieval outputs to structured, goal-directed reasoning. Extensive evaluation across mathematics, law, and healthcare domains—and across multiple LLMs—demonstrates average accuracy gains of 3–5%, reaching up to 7.5% in complex scenarios, while significantly improving interpretability and task adaptability.
📝 Abstract
The integration of external knowledge through Retrieval-Augmented Generation (RAG) has become foundational in enhancing large language models (LLMs) for knowledge-intensive tasks. However, existing RAG paradigms often overlook the cognitive step of applying knowledge, leaving a gap between retrieved facts and task-specific reasoning. In this work, we introduce RAG+, a principled and modular extension that explicitly incorporates application-aware reasoning into the RAG pipeline. RAG+ constructs a dual corpus consisting of knowledge and aligned application examples, created either manually or automatically, and retrieves both jointly during inference. This design enables LLMs not only to access relevant information but also to apply it within structured, goal-oriented reasoning processes. Experiments across mathematical, legal, and medical domains, conducted on multiple models, demonstrate that RAG+ consistently outperforms standard RAG variants, achieving average improvements of 3-5%, and peak gains up to 7.5% in complex scenarios. By bridging retrieval with actionable application, RAG+ advances a more cognitively grounded framework for knowledge integration, representing a step toward more interpretable and capable LLMs.