Information fusion strategy integrating pre-trained language model and contrastive learning for materials knowledge mining

2025-06-14

AI Summary

Modeling complex material properties—such as ductility in alloys—that are governed by intricate, coupled processing–microstructure–property relationships remains challenging due to the difficulty in quantitatively representing processing parameters and microstructural features. Method: This work proposes a novel multi-source heterogeneous information fusion framework that synergistically integrates textual semantics and physics-informed descriptors. It pioneers the coupling of MatSciBERT with contrastive learning to automatically extract implicit “processing–structure–property” knowledge from scientific literature, enabling joint embedding of domain-specific text and numerical features. Contribution/Results: Evaluated on a titanium alloy validation set and a refractory multi-principal-element alloy test set, the model achieves R² scores of 0.849 and 0.680, respectively—substantially outperforming conventional machine learning and purely data-driven approaches. By overcoming the bottleneck of quantitative characterization of processing-microstructure linkages, this study establishes a new paradigm for interpretable, knowledge-enhanced materials property prediction.

Technology Category

• Machine Learning: Matrix & Tensor Methods
• Data Mining & Knowledge Management: Intelligent Query Processing
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Machine learning has revolutionized materials design, yet predicting complex properties like alloy ductility remains challenging due to the influence of processing conditions and microstructural features that resist quantification through traditional reductionist approaches. Here, we present an innovative information fusion architecture that integrates domain-specific texts from materials science literature with quantitative physical descriptors to overcome these limitations. Our framework employs MatSciBERT for advanced textual comprehension and incorporates contrastive learning to automatically extract implicit knowledge regarding processing parameters and microstructural characteristics. Through rigorous ablation studies and comparative experiments, the model demonstrates superior performance, achieving coefficient of determination (R2) values of 0.849 and 0.680 on titanium alloy validation set and refractory multi-principal-element alloy test set. This systematic approach provides a holistic framework for property prediction in complex material systems where quantitative descriptors are incomplete and establishes a foundation for knowledge-guided materials design and informatics-driven materials discovery.

Problem

• Predicting alloy ductility with processing and microstructure factors
• Integrating textual and quantitative data for materials property prediction
• Overcoming incomplete descriptors in complex material systems

Innovation

• Integrates pre-trained language model MatSciBERT
• Uses contrastive learning for knowledge extraction
• Fuses textual and quantitative data for prediction