Aligning Language Models with Investor and Market Behavior for Financial Recommendations
Fernando Spadea and Oshani Seneviratne
大多数财务推荐系统往往无法解释关键的行为和监管因素,导致建议与用户偏好不一致,难以解释或不太可能遵循。 我们介绍了FLARKO(金融语言模型资产推荐与知识图谱优化),一个集成大型语言模型(LLM),知识图谱(KG)和卡尼曼 - 特弗斯基优化(KTO)的新框架,以生成既盈利又符合行为的资产建议。 FLARKO将用户的交易历史和资产趋势编码为结构化KG,为LLM提供可解释和可控的上下文。 为了证明我们方法的适应性,我们开发和评估集中架构(CenFLARKO)和联合变体(FedFLARKO)。 据我们所知,这是KTO首次展示将KTO用于微调LLM进行金融资产推荐。 我们还首次使用结构化KG来在联合学习(FL)环境中对行为金融数据进行LLM推理。 在FAR-Trans数据集上进行了评估,FLARKO在行为调整和联合盈利能力方面始终优于最先进的推荐基线,同时保持可解释和资源效率。
Most financial recommendation systems often fail to account for key behavioral and regulatory factors, leading to advice that is misaligned with user preferences, difficult to interpret, or unlikely to be followed. We present FLARKO (Financial Language-model for Asset Recommendation with Knowledge-graph Optimization), a novel framework that integrates Large Language Models (LLMs), Knowledge Graphs (KGs), and Kahneman-Tversky Optimization (KTO) to generate asset recommendations that are both prof...