![]() ![]() Our system obtained overall F-scores of 93.2% and 89.2% for these experiments, thus achieving state-of-the-art results.ConclusionsOur proposed representation learning approach leverages terminological embeddings to capture semantic similarity. We performed additional experiments on aligning FMA to NCI Thesaurus and to SNOMED CT based on a reference alignment extracted from the UMLS Metathesaurus. This compares favorably with the best performing systems on the Ontology Alignment Evaluation Initiative anatomy challenge. The resulting framework also incorporates a novel outlier detection mechanism based on a denoising autoencoder that is shown to improve performance.ResultsAn ontology matching system derived using the proposed framework achieved an F-score of 94% on an alignment scenario involving the Adult Mouse Anatomical Dictionary and the Foundational Model of Anatomy ontology (FMA) as targets. This embedding is derived on the basis of a novel phrase retrofitting strategy through which semantic similarity information becomes inscribed onto fields of pre-trained word vectors. Our approach is based on embedding ontological terms in a high-dimensional Euclidean space. Unlike past work that has focused on feature engineering, we present a novel representation learning approach that is tailored to the ontology matching task. Representation Learning For Ontology Matching:īackgroundWhile representation learning techniques have shown great promise in application to a number of different NLP tasks, they have had little impact on the problem of ontology matching.7 Sensitivity analysis of the proposed algorithm’s performance with different threshold values Table 6 Proposed algorithm’s performance in relation to the used synonymy information sourcesįig. 6 Correlation between the relative change in training data’s size and F1-score Importance Of The Ontology Extracted Synonymsįig.Table 5 Runtimes of the steps in the proposed algorithm Table 4 Sample misalignments produced by aligning ontologies using either SCBOW or Word2Vec vectors ![]() Table 3 Ablation study experiment’s listings 5 Feature ablation study of our proposed approach across all the experimental ontology matching tasks Table 2 Performance of ontology matching systems across the different matching tasks.įig. Table 1 Respective sizes of the ontology matching tasks 4 Overall proposed ontology matching architecture The input projection layer is omittedįig. 2 Phrase Retrofitting architecture based on a Siamese CBOW network and Knowledge Distillation. The dashed horizontal lines correspond to equivalence matchings between the NCI Thesaurus and the Mouse Anatomy ontologyįig. 1 Example of alignments between the NCI Thesaurus and the Mouse Ontology (adapted from ). Biomedical ontology alignment: an approach based on representation learning Authorsįig. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |