Ontology-aligned structuring and reuse of multimodal materials data and workflows towards automatic reproduction

TL;DR

The paper presents an ontology-driven, large language model–assisted workflow for automatically extracting, unifying, and reusing computational stacking fault energy workflows from Mg and Mg-based literature. By integrating CMSO, ASMO, PLDO ontologies with atomRDF and a canonical JSON schema, the approach enables semantically interoperable knowledge graphs that support systematic comparison of SFE values and the reuse of computational protocols. A multi-stage pipeline—comprising literature collection, targeted filtering, text/table extraction, and ontology alignment—addresses heterogeneity in reporting and aims to improve transparency and reproducibility in computational materials science. Limitations remain due to missing or implicit metadata, but the framework provides a concrete pathway toward post-publication data management and potentially automatic reproduction with human-in-the-loop guidance.

Abstract

Reproducibility of computational results remains a challenge in materials science, as simulation workflows and parameters are often reported only in unstructured text and tables. While literature data are valuable for validation and reuse, the lack of machine-readable workflow descriptions prevents large-scale curation and systematic comparison. Existing text-mining approaches are insufficient to extract complete computational workflows with their associated parameters. An ontology-driven, large language model (LLM)-assisted framework is introduced for the automated extraction and structuring of computational workflows from the literature. The approach focuses on density functional theory-based stacking fault energy (SFE) calculations in hexagonal close-packed magnesium and its binary alloys, and uses a multi-stage filtering strategy together with prompt-engineered LLM extraction applied to method sections and tables. Extracted information is unified into a canonical schema and aligned with established materials ontologies (CMSO, ASMO, and PLDO), enabling the construction of a knowledge graph using atomRDF. The resulting knowledge graph enables systematic comparison of reported SFE values and supports the structured reuse of computational protocols. While full computational reproducibility is still constrained by missing or implicit metadata, the framework provides a foundation for organizing and contextualizing published results in a semantically interoperable form, thereby improving transparency and reusability of computational materials data.

Figures (7)

• Figure 1: Overview of ontological SFE workflow description.
• Figure 2: Schematic of LLM-based data extraction and normalization workflow of SFE calculations.
• Figure 3: Decision-making process for automatic filtering.
• Figure 4: Hierarchical filtering and classification of the corpus in form of a Sankey diagram. The Sankey diagram illustrates the sequential application of abstract availability, study type, computational method, material system, compositional complexity, crystal structure, and SFE relevance criteria, with node widths proportional to the number of studies passing each stage. The height of each gray bar w.r.t. the left most purple bar corresponds to the fraction of paper retained after each filter.
• Figure 5: Distribution of section organization within the 200 papers of the dataset. IMRD is short for Introduction-Method-Results-Discussion.