🔬 研究主題與搜尋方向¶
這頁說明 arXiv 文獻自動化 是根據什麼研究主題與關鍵字在幫我找論文。
想調整方向:編輯 Research_Library/_profiles/llm-graphrag-goat.md 後重新發布(控制台「📋 更新研究主題頁」或發 profile 指令)。
研究題目:A Knowledge Graph-Enhanced Large Language Model Framework for Goat Disease Question Answering System
核心問題:GraphRAG 是否優於 Vector-only / LLM-only baseline 在領域特定獸醫 QA benchmark?
🔑 搜尋關鍵字(決定抓哪些論文)¶
必要(命中才當研究類)¶
標題/摘要至少要有一個
GraphRAG Graph-RAG graph-augmented knowledge graph graph retrieval graph rag
高度相關(命中加重分)¶
最想看到的方向
multi-hop retrieval multi-hop reasoning multi-hop QA hybrid retrieval hybrid search BM25 vector RRF reciprocal rank fusion cross-encoder reranking BGE reranker bidirectional retrieval entity resolution entity disambiguation knowledge graph QA Neo4j retrieval ablation Cypher triples extraction relation extraction transitive inference transitive multi-hop knowledge graph reasoning KG completion graph traversal reasoning inferential question answering multi-hop over knowledge graph implicit relation reasoning
中度相關¶
相關但非核心
RAG evaluation LLM judge LLM-as-judge DeepSeek-R1 vLLM retrieval-augmented generation dense retrieval sparse retrieval embedding model nomic-embed retrieval recall supernode graph traversal benchmark methodology golden chunks ablation study chain-of-thought retrieval multi-hop QA benchmark compositional reasoning
邊緣相關¶
其他方向也用得到
veterinary AI livestock LLM domain-specific QA prompt engineering retrieval chain-of-thought retrieval
🚫 排除(看到就降到 0 分)¶
明確不採用的方向
global GraphRAG community detection Leiden algorithm LLM finetuning LoRA fine-tuning chatbot UI 純向量檢索沒接 graph hiring / careers / interview tips
最後更新:2026-06-19 17:08