Recognizing lexical and semantic change patterns in evolving life science ontologies to inform mapping adaptation

Area of Science:

• Bioinformatics
• Ontology Engineering
• Life Sciences

Background:

• Life science ontologies evolve frequently, necessitating significant effort to maintain mappings.
• Automated methods for updating ontology mappings are crucial but lack detailed change descriptions.
• Current approaches often fail to capture nuanced concept evolution between ontology versions.

Purpose of the Study:

• To define and automatically recognize concept attribute change patterns in evolving ontologies.
• To evaluate the effectiveness of proposed methods for ontology mapping adaptation.
• To assess the impact of recognized change patterns on mapping strategies.

Main Methods:

• Developed methods to identify concept attribute change patterns based on attribute similarity.
• Proposed algorithms to automatically recognize these change patterns.
• Evaluated the performance of pattern recognition and its influence on mapping adaptation.

Main Results:

• Achieved Precision >60% and Recall >35% in identifying change patterns.
• Determined that recognized patterns impact ~66% of mapping adaptation actions positively.
• Demonstrated the influence of similarity coefficients on algorithm performance.

Conclusions:

• The approach effectively characterizes ontology evolution at the concept attribute level.
• Proposed change patterns are relevant for informed decision-making in ontology mapping adaptation.
• Experimental validation on real-world life science ontologies confirms the approach's efficacy.