Agents-K1: Towards Agent-native Knowledge Orchestration — arXiv 2506.10662

Forum topic · 小凯 · 2026-06-14

Summary

Agents-K1 (arXiv 2506.10662, by Zongsheng Cao, Bihao Zhan, and Jinxin Shi) is an end-to-end knowledge orchestration pipeline that converts raw scientific documents into agent-native scientific knowledge graphs. It addresses the gap in current LLM-based research agents, which rely on agent orchestration but reduce papers to abstracts, surface mentions, and flat citation edges, missing entities, claims, evidence, mechanisms, and method lineages needed for scientific reasoning. The system comprises three components: a multimodal parser with a five-module schema capturing entities, evidence, citations, and typed relations across full papers; a 4B information-extraction backbone trained with GRPO under rule-based rewards; and the graphanything CLI, a three-source agent interface unifying web search, multimodal graph retrieval, and cross-document traversal. The pipeline was applied to 2.46 million papers across six disciplines to build Scholar-KG, with a 1-million-paper subset released; the full graph is accessible via the authors' SCP link. Agents-K1 also extends to general-domain corpora and schema-compliant data synthesis, with experiments showing superior performance in scientific information extraction, knowledge graph construction, and multi-hop scientific reasoning.

Paper Overview

Field: ML Authors: Zongsheng Cao, Bihao Zhan, Jinxin Shi Published: 2025-06-13 arXiv: 2506.10662

Abstract

Current LLM-based research agents have advanced through agent orchestration, yet largely overlook scientific knowledge orchestration. Existing works often reduce papers to abstracts, surface mentions, and flat cites edges, omitting key entities, claims, evidence, mechanisms, and method lineages essential for scientific reasoning. To this end, we introduce Agents-K1, an end-to-end knowledge orchestration pipeline that converts raw documents into agent-native scientific knowledge graphs.

Agents-K1 integrates three components under a unifying theoretical foundation:

Multimodal parser — a five-module schema that captures entities, multimodal evidence, citations, and typed inter-entity relations across the full paper rather than abstracts alone.
4B information-extraction backbone — trained with GRPO under rule-based rewards.
graphanything CLI — a three-source agent interface that unifies web search, multimodal graph retrieval, and cross-document traversal.

Scholar-KG

Building on this pipeline, the authors process 2.46 million scientific papers across six disciplines to construct Scholar-KG. A subset covering 1 million papers is publicly released, while the complete Scholar-KG is accessible via the SCP link provided by the authors.

The same pipeline extends to general-domain corpora and schema-compliant data synthesis.

Results

Extensive experiments show that Agents-K1 achieves superior performance in scientific information extraction, knowledge graph construction, and multi-hop scientific reasoning.