LLMKB:基于知识库增强的大语言模型对话式推荐OA
LLMKB:Large Language Models with Knowledge Base Augmentation for Conversational Recommendation
会话推荐系统(conversational recommender system,CRS)侧重于通过自然语言交互和对话历史来细化偏好并提供个性化推荐.大语言模型(large language model,LLM)在各个领域都表现出了出色的性能,从而促使研究人员研究其在CRS中的适用性.然而,由于缺乏特定任务的知识和特征提取过程效率低下,LLM在CRS 任务中的表现仍然不是最佳的.因此,外部知识源,如知识图谱(knowledge graph,KG)和知识库(knowledge base,KB),通常被引入来解决数据稀疏性问题.与KG相比,KB具有更高的检索效率,使其更适用于将LLM作为推荐者的场景.为此,我们引入了一个将LLM与KB集成在一起的新框架,用于增强检索生成,即LLMKB.首先,LLMKB利用结构化知识来创建映射字典,从异构知识中提取实体关系信息以构建KB.其次,LLMKB通过检索模型微调来实现用户信息表示与KB中文档之间的嵌入校准.最后,LLMKB采用检索增强生成技术,基于融合的文本输入来生成推荐,并进行后处理.在两个公共 CRS 数据集上的实验结果证明了该框架的有效性.代码可公开获取:https://anonymous.4open.science/r/LLMKB-6FD0.
Conversational recommender systems(CRSs)focus on refining preferences and providing personalized recommendations through natural language interactions and dialogue history.Large language models(LLMs)have shown outstanding performance across various domains,thereby prompting researchers to investigate their applicability in recommendation systems.However,due to the lack of task-specific knowledge and an inefficient feature extraction process,LLMs still have suboptimal performance in recommendation tasks.Therefore,external knowledge sources,such as knowledge graphs(KGs)and knowledge bases(KBs),are often introduced to address the issue of data sparsity.Compared to KGs,KBs possess higher retrieval efficiency,making them more suitable for scenarios where LLMs serve as recommenders.To this end,we introduce a novel framework integrating LLMs with KBs for enhanced retrieval generation,namely LLMKB.LLMKB initially leverages structured knowledge to create mapping dictionaries,extracting entity-relation information from heterogeneous knowledge to construct KBs.Then,LLMKB achieves the embedding calibration between user information representations and documents in KBs through retrieval model fine-tuning.Finally,LLMKB employs retrieval-augmented generation to produce recommendations based on fused text inputs,followed by post-processing.Experiment results on two public CRS datasets demonstrate the effectiveness of our framework.Our code is publicly available at the link:https://anonymous.4open.science/r/LLMKB-6FD0.
方秀;邱斯嘉;孙国豪;陆金虎
东华大学 计算机科学与技术学院,上海 201620东华大学 计算机科学与技术学院,上海 201620东华大学 计算机科学与技术学院,上海 201620东华大学 计算机科学与技术学院,上海 201620
信息技术与安全科学
推荐系统大语言模型(LLM)知识库(KB)
recommender systemlarge language model(LLM)knowledge base(KB)
《东华大学学报(英文版)》 2026 (1)
91-103,13
评论