LAUSR

Abstract Due to high-throughput experiments and computations, data-generating is relatively easier than before. Currently, materials digital data gradually become one of the key factors of data-driven based novel materials design. Therefore, how to extract useful knowledge from the massive data is a hot topic in materials informatics. In this paper, ontology is used to build a knowledge base, and then production rules and an inference machine are used to derive new knowledge. To realize from the database to the knowledge base of circulation, and then extract the knowledge, further enrich the knowledge base. In experiments, we use stainless steel as an example. And production rules are used to represent the domain knowledge according to the influence of stainless steel metallographic structure and chemical composition on corrosion resistance. And the experimental results show that when adding different rules to the inference engine for reasoning, the knowledge number increases from 202 to 264, 326, 388, 450, 512, 518 and 642 respectively. Moreover, after knowledge reasoning, materials corrosion resistances are added to the knowledge base, including resistance to atmospheric corrosion, acid resistance, stress corrosion crack resistance, resistance to hole corrosion and clearance corrosion, low-temperature strength and toughness and room temperature strength. Therefore, using production-based reasoning, the metallographic structure, corrosion resistance, and materials composition are semantically correlated that can hence enrich the knowledge from the existing data.