Bridging Tool Dependencies and Domain Knowledge: A Graph-Based Framework for In-Context Planning

TL;DR

The paper tackles the cold-start problem in in-context planning for domain-specific tools by constructing a tool knowledge graph from tool schemas and fusing it with a domain knowledge graph derived from internal documents. It introduces a Deep Research-inspired dependency extraction pipeline and GraphRAG-based domain graph construction, followed by Neptune-based graph fusion and a HippoRAG2-inspired dense-sparse integration to generate exemplar plans. Experiments on ToolBench demonstrate strong dependency-detection performance across multiple LLMs and improved exemplar plan generation when the fused graph is used, with ablations highlighting the value of Personalized PageRank. The approach enables robust tool-augmented reasoning and planning in enterprise settings by bridging tool interactions with procedural knowledge.

Abstract

We present a framework for uncovering and exploiting dependencies among tools and documents to enhance exemplar artifact generation. Our method begins by constructing a tool knowledge graph from tool schemas,including descriptions, arguments, and output payloads, using a DeepResearch-inspired analysis. In parallel, we derive a complementary knowledge graph from internal documents and SOPs, which is then fused with the tool graph. To generate exemplar plans, we adopt a deep-sparse integration strategy that aligns structural tool dependencies with procedural knowledge. Experiments demonstrate that this unified framework effectively models tool interactions and improves plan generation, underscoring the benefits of linking tool graphs with domain knowledge graphs for tool-augmented reasoning and planning.

Figures (2)

• Figure 1: Tool Graph Construction
• Figure 2: Dense-Sparse Integration framework for exemplar plans generation. Red dots represent document nodes; blue dots represent tool nodes.