Abstract:With the rapid development of large-scale language models, Retrieval-Augmented Generation (RAG) has been widely adopted. However, existing RAG paradigms are inevitably influenced by erroneous retrieval information, thereby reducing the reliability and correctness of generated results. Therefore, to improve the relevance of retrieval information, this study proposes a method that replaces traditional retrievers with GPT-3.5, leveraging its vast corpus knowledge to generate retrieval information. We also propose a web retrieval based method to implement fine-grained knowledge retrieval, Utilizing the powerful reasoning capability of GPT-3.5 to realize semantic partitioning of <a class="link-external link-http" href="http://problem.In" rel="external noopener nofollow">this http URL</a> order to mitigate the illusion of GPT retrieval and reduce noise in Web retrieval,we proposes a multi-source retrieval framework, named MSRAG, which combines GPT retrieval with web retrieval. Experiments on multiple knowledge-intensive QA datasets demonstrate that the proposed framework in this study performs better than existing RAG framework in enhancing the overall efficiency and accuracy of QA systems.

What problem does this paper attempt to address?

The aim of this paper is to address the issues of information relevance and noise in traditional Retrieval-Augmented Generation (RAG) frameworks. Existing RAG methods may be affected by erroneous retrieval information, which lowers the reliability and accuracy of the generated results. Therefore, this paper proposes a new framework called MSRAG (Multi-Source Retrieval Augmented Generation). First, MSRAG replaces the traditional retriever with GPT-3.5 to generate retrieval information using its vast corpus knowledge in order to improve information relevance. Secondly, fine-grained knowledge retrieval is achieved through a web-based retrieval approach, utilizing GPT-3.5's powerful reasoning capability for semantic segmentation. To mitigate the hallucination effect of GPT retrieval and reduce noise from web retrieval, MSRAG combines GPT retrieval with web retrieval, forming a multi-source retrieval framework. Experiments on multiple knowledge-intensive question answering datasets verify the effectiveness of MSRAG, demonstrating its superiority over other RAG frameworks in enhancing the overall efficiency and accuracy of question answering systems. The specific contributions are as follows: 1. Use GPT-3.5 for retrieval to improve information relevance and achieve semantic segmentation of queries through a web retrieval framework. 2. Propose a multi-source retrieval method that combines GPT retrieval with web retrieval to reduce erroneous information and noise. 3. Experimental results on different datasets demonstrate the outstanding performance of MSRAG in improving task performance for question answering. The paper also conducts ablation studies to validate the effectiveness of GPT retrieval and the advantages of the integrated MSRAG framework. Future research may focus on improving the performance of GPT retrieval and exploring methods to reduce the operational cost and accelerate the runtime of the integrated approach.

A Multi-Source Retrieval Question Answering Framework Based on RAG

Towards Multi-Source Retrieval-Augmented Generation via Synergizing Reasoning and Preference-Driven Retrieval

Hierarchical Retrieval-Augmented Generation Model with Rethink for Multi-hop Question Answering

DR-RAG: Applying Dynamic Document Relevance to Retrieval-Augmented Generation for Question-Answering

Retrieval-Augmented Generation for Domain-Specific Question Answering: A Case Study on Pittsburgh and CMU

EfficientRAG: Efficient Retriever for Multi-Hop Question Answering

LongRAG: A Dual-Perspective Retrieval-Augmented Generation Paradigm for Long-Context Question Answering

Enhancing Retrieval and Managing Retrieval: A Four-Module Synergy for Improved Quality and Efficiency in RAG Systems

REAR: A Relevance-Aware Retrieval-Augmented Framework for Open-Domain Question Answering

RuleRAG: Rule-guided retrieval-augmented generation with language models for question answering

HiQA: A Hierarchical Contextual Augmentation RAG for Multi-Documents QA

Layered Query Retrieval: an Adaptive Framework for Retrieval-Augmented Generation in Complex Question Answering for Large Language Models

MBA-RAG: a Bandit Approach for Adaptive Retrieval-Augmented Generation through Question Complexity

Retriever-and-Memory: Towards Adaptive Note-Enhanced Retrieval-Augmented Generation

RichRAG: Crafting Rich Responses for Multi-faceted Queries in Retrieval-Augmented Generation

WeKnow-RAG: An Adaptive Approach for Retrieval-Augmented Generation Integrating Web Search and Knowledge Graphs

Retrieval-Augmented Generation for Large Language Models: A Survey

FoRAG: Factuality-optimized Retrieval Augmented Generation for Web-enhanced Long-form Question Answering

Retrieve, Summarize, Plan: Advancing Multi-hop Question Answering with an Iterative Approach

W-RAG: Weakly Supervised Dense Retrieval in RAG for Open-domain Question Answering

MAIN-RAG: Multi-Agent Filtering Retrieval-Augmented Generation