Track synthesis, materials and catalysis as it is published.
Search papers, labs, and topics across Lattice.
This paper introduces SciKGExtract, a schema-guided framework that leverages LLM extraction, chemical normalization, and agent-based refinement to overcome the heterogeneity of materials literature for knowledge graph integration. By employing PubChem normalization and a multi-agent evaluation loop, the framework significantly improves the F1 score for extracting experimental details from scientific papers, demonstrating a substantial leap from direct extraction to agentically refined knowledge. The evaluation on atomic-layer-deposition papers for ZnO and IGZO highlights the framework's efficacy, particularly the marked improvement for ZnO (0.591 to 0.805 F1), while also exposing the challenges of complex multicomponent processes.
Agentic refinement boosts materials process knowledge extraction F1 by 36% for ZnO, revealing critical errors in process segmentation and numerical assignment within complex scientific literature.
Materials literature contains detailed experimental knowledge, but procedures, chemical entities and measurements remain difficult to aggregate because they are reported in heterogeneous forms and depend on process-specific context. We present SciKGExtract, a schema-guided framework that combines large-language-model extraction with chemical normalization and agent-based evaluation and refinement before knowledge-graph integration. We evaluate the framework on 176 atomic-layer-deposition papers describing zinc oxide (ZnO) and indium--gallium--zinc oxide (IGZO), together with an expert-annotated full-schema subset. PubChem normalization improves exact-match extraction F1 for every tested model. For ZnO, the best F1 increases from 0.591 for direct normalized extraction to 0.805 with agentic refinement, whereas the best IGZO result is 0.344, revealing the greater difficulty of multicomponent supercycle processes. Evaluation against a deeply nested schema containing 65 experimental properties and 155 quantitative measurement nodes further exposes errors in process segmentation and numerical assignment. These results show that chemical canonicalization and targeted agentic verification provide complementary controls for converting complex materials literature into reusable, machine-actionable experimental knowledge.