TAPS: Tool-Augmented Personalisation via Structured Tagging
This addresses the challenge of personalising tool use in dialogue agents for users, though it appears incremental as it builds on existing tool-augmented LLM approaches.
The paper tackled the problem of integrating user preferences into tool-augmented large language models for goal-oriented dialogue agents, and the result was that TAPS significantly improved personalisation, achieving state-of-the-art performance for open-source models on the NLSI task.
Recent advancements in tool-augmented large language models have enabled them to interact with external tools, enhancing their ability to perform complex user tasks. However, existing approaches overlook the role of personalisation in guiding tool use. This work investigates how user preferences can be effectively integrated into goal-oriented dialogue agents. Through extensive analysis, we identify key weaknesses in the ability of LLMs to personalise tool use. To this end, we introduce TAPS, a novel solution that enhances personalised tool use by leveraging a structured tagging tool and an uncertainty-based tool detector. TAPS significantly improves the ability of LLMs to incorporate user preferences, achieving the new state-of-the-art for open source models on the NLSI task.