Hybridizing Fuzzy String Matching and Machine Learning for Improved Ontology Alignment

Ontology alignment has become an important process for identifying similarities and differences between ontologies, to facilitate their integration reuse. To this end, fuzzy string-matching algorithms have been developed strings similarity detection used in ontology alignment. However, a significant limitation of existing is reliance on lexical/syntactic contents only, which do not capture semantic features ontologies. address limitation, paper proposed novel method that hybridizes the Deep Bidirectional Transformer (BERT) deep learning model with three machine regression classifiers, namely, K-Nearest Neighbor Regression (kNN), Decision Tree (DTR), Support Vector (SVR), perform The use kNN, SVR, DTR classifiers resulted building models (SM), encoded SM-kNN, SM-SVR, SM-DTR, respectively. experiments were conducted dataset obtained from anatomy track Alignment Evaluation Initiative 2022 (OAEI 2022). performances SM-DTR evaluated using various metrics including precision, recall, F1-score, accuracy at thresholds 0.70, 0.80, 0.90, as well error rates running times. experimental results revealed SM-SVR achieved best recall 1.0, while exhibited accuracy, F1-score 0.98, 0.97, Furthermore, showed outperformed state-of-the-art systems participated OAEI challenge, indicating superior capability method.