Contextual Semantic Embeddings for Ontology Subsumption Prediction

Chen, Jiaoyan; He, Yejun; Jiménez-Ruiz, Ernesto; Dong, Hua; Horrocks, Ian

doi:10.48550/arxiv.2202.09791

Cited by 6 publications

(10 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Such vectors can then be used as additional inputs to machine learning models based on neural networks, to improve their performance by allowing them to use an approximation of expert knowledge [87]. A good method of learning embeddings from symbolic knowledge should ideally leverage the structural information present in relations between abstract concepts, and should not try to learn embeddings for abstract concepts with the aid of word embeddings, due to the ambiguity of language, its limited abstraction, and other problems [11].…”

Section: Knowledge Base Embeddingsmentioning

confidence: 99%

“…In general, concept embeddings are represented by vectors in real vector space R N e , where N e is the embedding dimension. In Equations ( 5)- (11) we define the recursive function h K : ALC → R N e that maps ALC concepts for knowledge base K to embedding vectors in real vector space R N e . We begin with the embedding for each concept name A i : a vector W K,A i of dimension N e , as given by Equation (5).…”

Section: Embedding Layermentioning

confidence: 99%

“…To extend h K to concept unions and universal restrictions, we use the de Morgan's laws [78], arriving at the formulas given in Equation (11).…”

Section: Embedding Layermentioning

confidence: 99%

See 2 more Smart Citations

Reason-able embeddings: Learning concept embeddings with a transferable neural reasoner

Adamski

Potoniec

2024

View full text Add to dashboard Cite

We present a novel approach for learning embeddings of ALC knowledge base concepts. The embeddings reflect the semantics of the concepts in such a way that it is possible to compute an embedding of a complex concept from the embeddings of its parts by using appropriate neural constructors. Embeddings for different knowledge bases are vectors in a shared vector space, shaped in such a way that approximate subsumption checking for arbitrarily complex concepts can be done by the same neural network, called a reasoner head, for all the knowledge bases. To underline this unique property of enabling reasoning directly on embeddings, we call them reason-able embeddings. We report the results of experimental evaluation showing that the difference in reasoning performance between training a separate reasoner head for each ontology and using a shared reasoner head, is negligible.

show abstract

Section: Knowledge Base Embeddingsmentioning

confidence: 99%

Section: Embedding Layermentioning

confidence: 99%

See 1 more Smart Citation

Reason-able embeddings: Learning concept embeddings with a transferable neural reasoner

Adamski

Potoniec

2024

View full text Add to dashboard Cite

show abstract

“…BERTSubs with Isolated Class (IC). BERTSubs with the IC setting [4] has the same architecture as BERTMap, but it fine-tunes the BERT model by the declared subsumptions in the two ontologies for matching. The current results are based on the labels defined by rdfs:label.…”

Section: Subsumption Matchingmentioning

confidence: 99%

“…However, there are often more subsumption mappings than equivalence mappings between real-world ontologies, and the former could play an important role in knowledge integration and ontology curation. With the blooming research and application of ML and text understanding techniques, systems for subsumption matching (e.g., BERTSubs [4]) will likely become more feasible and widely investigated. 4.…”

Section: Introductionmentioning

confidence: 99%

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

Chen

Dong

et al. 2022

The Semantic Web – ISWC 2022

View full text Add to dashboard Cite

Ontology Matching (OM) plays an important role in many domains such as bioinformatics and the Semantic Web, and its research is becoming increasingly popular, especially with the application of machine learning (ML) techniques. Although the Ontology Alignment Evaluation Initiative (OAEI) represents an impressive effort for the systematic evaluation of OM systems, it still suffers from several limitations including limited evaluation of subsumption mappings, suboptimal reference mappings, and limited support for the evaluation of ML-based systems. To tackle these limitations, we introduce five new biomedical OM tasks involving ontologies extracted from Mondo and UMLS. Each task includes both equivalence and subsumption matching; the quality of reference mappings is ensured by human curation, ontology pruning, etc.; and a comprehensive evaluation framework is proposed to measure OM performance from various perspectives for both ML-based and non-ML-based OM systems. We report evaluation results for OM systems of different types to demonstrate the usage of these resources, all of which are publicly available as part of the new Bio-ML track at OAEI 2022.

show abstract

HURON: A Quantitative Framework for Assessing Human Readability in Ontologies

Abad-Navarro,

Martínez-Costa,

Fernández-Breis

2023

IEEE Access

View full text Add to dashboard Cite

The increasing use of ontologies requires their quality assurance. Ontology quality assurance consists of a set of activities that allow analyzing the ontology, identifying strengths and weaknesses, and proposing improvement actions. Human readability is a quality aspect that improves the use and reuse of ontologies. The human readable content refers to the natural language content consumed by humans and by the growing number of embedding methods applied to ontologies. The ontology community has proposed best practices related to human readability, but there is no standardized framework for its assessment. We aim to provide a framework for analyzing the human readability based on quantitative metrics for supporting ontology developers' decisions. We present the HURON framework, which consists of the specification of five quantitative metrics related to the human readability of ontology content, and a software tool to implement them. The metrics take into account the number of names, descriptions, or synonyms, and also assess the application of systematic naming conventions and the 'lexically suggest, lexically define' principle. Target values are provided for each metric to help to interpret the values. HURON can also be used to assess compliance with best practices. We have applied HURON to a representative set of biomedical ontologies, the OBO Foundry repository. The results showed that, in general, the OBO Foundry ontologies comply with the expected number of descriptions and names in their classes, and both lexical and semantically formalized contents are aligned. However, most of the ontologies did not follow a systematic naming convention. In general, the ontologies of this repository show adherence to some of the best practices, although areas of improvement were identified. A number of recommendations are made for ontology developers and users. INDEX TERMS knowledge engineering, ontologies, quality assurance, readability metrics, semantic web I. INTRODUCTIONOntologies play a key role in knowledge engineering by providing a common conceptualization of a domain. Ontologies have been successfully applied in different domains, but especially in biology and biomedicine, with different purposes [1][2][3][4][5].At the time of writing, repositories like BioPortal [6] had more than 1,000 ontologies and both the Open Biological and Biomedical Ontology (OBO) Foundry [7] and the Ontology Lookup Service (OLS) [8] had more than 250 ontologies. The number of ontologies in these repositories is continuously increasing, which demonstrates their relevance and impact.In contrast to other artifacts used in data management systems, such as relational databases, developed specifically for particular applications, ontologies should be created in a standardized way to facilitate their reuse. The sharing and reuse orientation of ontologies have also made them fundamental for obtaining Findable, Accessible, Interoperable, and Reusable (FAIR) datasets [9]. As a consequence, assuring the quality of ontologies has become a major need.

show abstract

Contextual Semantic Embeddings for Ontology Subsumption Prediction

Cited by 6 publications

References 11 publications

Reason-able embeddings: Learning concept embeddings with a transferable neural reasoner

Reason-able embeddings: Learning concept embeddings with a transferable neural reasoner

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

HURON: A Quantitative Framework for Assessing Human Readability in Ontologies

Contact Info

Product

Resources

About