Improving Complex Reasoning over Knowledge Graph with Logic-Aware Curriculum Tuning
This work addresses the challenge of complex reasoning for knowledge graph applications, representing an incremental improvement by integrating LLMs with curriculum learning.
The paper tackled the problem of answering complex queries over incomplete knowledge graphs by proposing a logic-aware curriculum tuning framework (LACT) based on large language models, which achieved an average +5.5% MRR score improvement over advanced methods, setting a new state-of-the-art.
Answering complex queries over incomplete knowledge graphs (KGs) is a challenging job. Most previous works have focused on learning entity/relation embeddings and simulating first-order logic operators with various neural networks. However, they are bottlenecked by the inability to share world knowledge to improve logical reasoning, thus resulting in suboptimal performance. In this paper, we propose a complex reasoning schema over KG upon large language models (LLMs), containing a curriculum-based logical-aware instruction tuning framework, named LACT. Specifically, we augment the arbitrary first-order logical queries via binary tree decomposition, to stimulate the reasoning capability of LLMs. To address the difficulty gap among different types of complex queries, we design a simple and flexible logic-aware curriculum learning framework. Experiments across widely used datasets demonstrate that LACT has substantial improvements~(brings an average +5.5% MRR score) over advanced methods, achieving the new state-of-the-art.