CCL-CM: cognitive-concept learning and conflict monitoring integration

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Explainable models based on concept learning offer relatively intuitive explanations for decisions and hold considerable theoretical appeal. In practice, particularly in classification tasks, they often face challenges related to accuracy and generalization. Drawing inspiration from cognitive conflict theory, we propose a novel optimization methodology that integrates concept learning with conflict detection mechanisms. By incorporating the spatial representation framework from cognitive space theory, we optimize the parameter adjustment process in concept learning models. This approach enhances the model’s cognitive behavior and strengthens its concept learning capabilities through the integration of multimodal information. In addition, conflict monitoring serves as a feedback mechanism that helps the model adjust its learning strategy when errors or inconsistencies are detected, thereby improving the generalization performance of interpretable models. We validate this method on four datasets, achieving higher classification accuracy.