Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces a knowledge graph-based external memory framework, extending the AriGraph architecture with hybrid edges and diverse retrieval mechanisms (A*, WaterCircles, beam search), to enhance LLM personalization. The framework dynamically constructs and updates the knowledge graph using the LLM itself, enabling structured memory for long-term interactions. Evaluations on TriviaQA, HotpotQA, and a temporally-extended DiaASQ demonstrate that optimal memory/retrieval configurations vary by task, and the system maintains robustness even with temporal dependencies and contradictory statements.
LLMs can now leverage dynamically constructed knowledge graphs with hybrid edges and tailored retrieval strategies to achieve robust personalization and context-aware reasoning in long-term interactions.
Personalizing language models by effectively incorporating user interaction history remains a central challenge in the development of adaptive AI systems. While large language models (LLMs), combined with Retrieval-Augmented Generation (RAG), have improved factual accuracy, they often lack structured memory and fail to scale in complex, long-term interactions. To address this, we propose a flexible external memory framework based on a knowledge graph that is constructed and updated automatically by the LLM. Building upon the AriGraph architecture, we introduce a novel hybrid graph design that supports both standard edges and two types of hyper-edges, enabling rich and dynamic semantic and temporal representations. Our framework also supports diverse retrieval mechanisms, including A*, WaterCircles traversal, beam search, and hybrid methods, making it adaptable to different datasets and LLM capacities. We evaluate our system on TriviaQA, HotpotQA, DiaASQ benchmarks and demonstrate that different memory and retrieval configurations yield optimal performance depending on the task. Additionally, we extend the DiaASQ benchmark with temporal annotations and internally contradictory statements, showing that our system remains robust and effective in managing temporal dependencies and context-aware reasoning.