Injecting Domain-Specific Knowledge into Large Language Models: A Comprehensive Survey

• DOI: 10.48550/arxiv.2502.10708
• Publication Date: 2025-02-15
• Authors: Zhihua Song; Bin Yan; Yuhan Liu; Miao Fang; Mingzhe Li; Rui Yan; Xiuying Chen
• Keywords: FOS: Computer and information sciences; Computer Science - Computation and Language; Computation and Language (cs.CL)

Large Language Models (LLMs) have demonstrated remarkable success in various tasks such as natural language understanding, text summarization, and machine translation. However, their general-purpose nature often limits effectiveness domain-specific applications that require specialized knowledge, healthcare, chemistry, or legal analysis. To address this, researchers explored diverse methods to enhance LLMs by integrating knowledge. In this survey, we provide a comprehensive overview of these methods, which categorize into four key approaches: dynamic knowledge injection, static embedding, modular adapters, prompt optimization. Each approach offers unique mechanisms equip with domain expertise, balancing trade-offs between flexibility, scalability, efficiency. We discuss how enable tackle tasks, compare advantages disadvantages, evaluate against general LLMs, highlight the challenges opportunities emerging field. For those interested delving deeper area, also summarize commonly used datasets benchmarks. keep updated on latest studies, maintain an open-source at: https://github.com/abilliyb/Knowledge_Injection_Survey_Papers, dedicated documenting research field LLM.