Machine Learning-Friendly Biomedical Datasets for Equivalence and Subsumption Ontology Matching

He, Yuan; Chen, Jiaoyan; Dong, Hang; Jiménez-Ruiz, Ernesto; Hadian, Ali; Horrocks, Ian

doi:10.1007/978-3-031-19433-7_33

Cited by 9 publications

(4 citation statements)

References 27 publications

(40 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, in this study, BERT is combined with fuzzy string-matching algorithms in a fully automated way for ontology alignment. In the study [42], the authors utilized the BERTMap system introduced by He et al [40] to perform the alignment of biomedical ontologies and concluded that BERTMap is convenient for real-world applications. In another study by Bajaj et al [43], the authors study the BERT biomedical variants to see whether they outperform the Siamese Network and original BioBERT.…”

Section: Related Workmentioning

confidence: 99%

Hybridizing Fuzzy String Matching and Machine Learning for Improved Ontology Alignment

Rudwan

Fonou-Dombeu

2023

Future Internet

View full text Add to dashboard Cite

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

show abstract

Section: Related Workmentioning

confidence: 99%

Hybridizing Fuzzy String Matching and Machine Learning for Improved Ontology Alignment

Rudwan

Fonou-Dombeu

2023

Future Internet

View full text Add to dashboard Cite

show abstract

“…Ontology Pruning We randomly sample a portion (10% or 20%) of the entities in the target ontology, remove them and link the parent and child of each removed entity to preserve the hierarchy as in He et al (2022). This forms a new ontology.…”

Section: Datasetsmentioning

confidence: 99%

Reveal the Unknown: Out-of-Knowledge-Base Mention Discovery with Entity Linking

Dong¹,

Chen²,

He³

et al. 2023

Preprint

View full text Add to dashboard Cite

Discovering entity mentions that are out of a Knowledge Base (KB) from texts plays a critical role in KB maintenance, but has not yet been fully explored. The current methods are mostly limited to the simple thresholdbased approach and feature-based classification; the datasets for evaluation are relatively rare. In this work, we propose BLINKout, a new BERT-based Entity Linking (EL) method which can identify mentions that do not have a corresponding KB entity by matching them to a special NIL entity. To this end, we integrate novel techniques including NIL representation, NIL classification, and synonym enhancement. We also propose Ontology Pruning and Versioning strategies to construct out-of-KB mentions from normal, in-KB EL datasets. Results on four datasets of clinical notes and publications show that BLINKout outperforms existing methods to detect out-of-KB mentions for medical ontologies UMLS and SNOMED CT.

show abstract

“…The closest related work is from Wang et al [29]. They apply LLaMa 65B [26], GPT3.5, and GPT4 to the Biomedical Datasets for Equivalence and Subsumption Matching [5]. The candidate generation is done by computing top k neighbors in an embedding space generated out of SapBERT [13] (a pre-trained BERT model designed for the biomedical domain).…”

Section: Large Language Models For Ontology Matchingmentioning

confidence: 99%

“…The high-precision matcher is a simple matcher in MELT that efficiently searches for concepts with the exact same normalized label (or URI fragment if a label is not available). 5 The normalization includes lowercasing, camel case, and deletion of non alpha-numeric characters. If there is only one such candidate for a concept, then it is matched.…”

Section: High-precision Matchermentioning

confidence: 99%

OLaLa: Ontology Matching with Large Language Models

Hertling,

Paulheim

2023

Proceedings of the 12th Knowledge Capture Conference 2023

View full text Add to dashboard Cite

Ontology (and more generally: Knowledge Graph) Matching is a challenging task where information in natural language is one of the most important signals to process. With the rise of Large Language Models, it is possible to incorporate this knowledge in a better way into the matching pipeline. A number of decisions still need to be taken, e.g., how to generate a prompt that is useful to the model, how information in the KG can be formulated in prompts, which Large Language Model to choose, how to provide existing correspondences to the model, how to generate candidates, etc. In this paper, we present a prototype that explores these questions by applying zero-shot and few-shot prompting with multiple open Large Language Models to different tasks of the Ontology Alignment Evaluation Initiative (OAEI). We show that with only a handful of examples and a well-designed prompt, it is possible to achieve results that are en par with supervised matching systems which use a much larger portion of the ground truth.

show abstract

Machine Learning-Friendly Biomedical Datasets for Equivalence and Subsumption Ontology Matching

Cited by 9 publications

References 27 publications

Hybridizing Fuzzy String Matching and Machine Learning for Improved Ontology Alignment

Hybridizing Fuzzy String Matching and Machine Learning for Improved Ontology Alignment

Reveal the Unknown: Out-of-Knowledge-Base Mention Discovery with Entity Linking

OLaLa: Ontology Matching with Large Language Models

Contact Info

Product

Resources

About