Recursive approach to combine expert knowledge and data-driven RSW weldability certification decision making process

Abstract

• Recursive and collaborative approach where knowledge gained from machine learning models is integrated with ontological knowledge. • Ontology-based semantic knowledge framework supports recursive communication with experts for data-driven RST weldability certification. • Extracted RSW concepts from the decision trees formalized by the RSW ontology and converted the decision rules into SWRL rules. • Transformed datasets helped to develop improved machine learning models that work as a new source of weldability prediction. Data-driven techniques have shown promising results in the analysis and understanding of complex welding processes. Data analytics play a significant role to turn data into valuable insights to assist in the weldability certification decision-making for Resistance Spot Welding (RSW) as well. However, to successfully perform the associated data analytics, domain knowledge is essential to construct more ‘sense-making’ analytics models, as often the models cannot properly capture the nuances of the domain and do not properly indicate the relationship among the RSW concepts and parameters. Thus, machine learning models developed from rough experimental data often do not provide models meaningful and sensible to the domain expert. In this article, we employ a recursive approach between the domain experts and data-driven models so that the knowledge of the domain experts can be integrated into the weldability certification decision-making process. An ontology-based semantic knowledge framework supports this recursive communication while helping the experts to instil more confidence in the developed analytics models. The collaborative and recursive approach implemented in this study helps the domain experts to tap into their domain knowledge and form expert opinions using the formalized semantic RSW concepts and decision rules. The expert opinions are then used to learn new knowledge about the RSW domain and transform the RSW datasets by incorporating significant features that were not included in the earlier models. The transformed datasets help us to develop improved machine learning models, which in turn work as a new source of semantic knowledge, as we have discovered through our pilot implementation.