Ontology alignment using artificial neural network for large-scale ontologies

Djeddi, Warith Eddine; Khadir, Mohamed Tarek

doi:10.1504/ijmso.2013.054180

“…However it will still be interesting to view the MLNK semantic, in a manner where a formal semantic is directly constructed on the structure of the MLNK, and propose then a correct and complete reasoning algorithm better adapted to the structure. On the other hand, an extension of the XM AP ++ alignment tool proposed in [10] Table 7 Multi-level networked knowledge example for impact level and typologies alignments over different MLNK levels as well as semiautomatic insertion of links , is planned. Consider the possibility of exploiting alignments at a certain level may help discovering alignments on a superior one, and improve alignment techniques.…”

Section: Discussionmentioning

confidence: 99%

Multi-level networked knowledge: representation and DL-reasoning

Klai

¹

,

Zimmermann

²

,

Khadir

³

2016

IJMSO

Self Cite

View full text Add to dashboard Cite

International audienceIntegrating pre-existing, heterogeneous, and complementary ontologies, and exploiting them jointly in reasoning remains a major challenge. Ontology alignments make explicit the correspondences between terms from different ontologies and must be taken into account in reasoning. Two forms of correspondences can be introduced: mappings represent predefined relations such as subsumption, equivalence, or disjointness, that have a fixed semantics in all interpretations; links can relate complementary ontologies by introducing terms defined by experts, and their semantics varies according to interpretations. Different experts can introduce different terms according to their points of view, which brings semantically heterogeneous links. Thus, integrating pre-existing networks of aligned ontologies requires aligning terminologies from different alignments, so as to form higher level alignments. This generates networked knowledge that can in turn be aligned with other networked knowledge. As a result, we talk of multi-level networked knowledge, a concept that we formalise here and for which we propose a possible formal semantic for automating reasoning tasks. This semantic consists of reducing reasoning on networked knowledge to reasoning over DL formalisms for which we have reasoning procedures. The proposed approach is implemented and tested in order to compare results, for different networks

show abstract

“…The use of supervised machine learning is being increasingly applied for a simplified configuration of schema and ontology matching, since it can be considered a way to aggregate several similarity metrics or matchers, removing the need to set manual thresholds or use vector distance metrics such as the cosine similarity, which give the same weight to all features [8], [9], [12], [17], [21], [22], [32], [32], [34].…”

Section: Mypriceindiacommentioning

confidence: 99%

Towards the smart use of embedding and instance features for property matching

Ayala

¹

,

Hernández

²

,

Ruiz

³

et al. 2021

2021 IEEE 37th International Conference on Data Engineering (ICDE)

0

View full text Add to dashboard Cite

Data integration tasks such as the creation and extension of knowledge graphs involve the fusion of heterogeneous entities from many sources. Matching and fusion of such entities require to also match and combine their properties (attributes). However, previous schema matching approaches mostly focus on two sources only and often rely on simple similarity measurements. They thus face problems in challenging use cases such as the integration of heterogeneous product entities from many sources.We therefore present a new machine learning-based property matching approach called LEAPME (LEArning-based Property Matching with Embeddings) that utilizes numerous features of both property names and instance values. The approach heavily makes use of word embeddings to better utilize the domain-specific semantics of both property names and instance values. The use of supervised machine learning helps exploit the predictive power of word embeddings.Our comparative evaluation against five baselines for several multi-source datasets with real-world data shows the high effectiveness of LEAPME.

show abstract

“…The use of supervised machine learning is being increasingly applied for a simplified configuration of schema and ontology matching, since it can be considered a way to aggregate several similarity metrics or matchers, removing the need to set manual thresholds or use vector distance metrics such as the cosine similarity, which give the same weight to all features [28,29,30,31,32,33,30,34,35]. The training data consists of the similarity of matching and non-matching pairs of schema/ontology elements together with multiple similarity values, e.g., according to different linguistic and structural similarities.…”

Section: Related Workmentioning

confidence: 99%

LEAPME: Learning-based Property Matching with Embeddings

Ayala¹,

Hernández²,

Ruiz³

et al. 2020

Preprint

0

View full text Add to dashboard Cite

Data integration tasks such as the creation and extension of knowledge graphs involve the fusion of heterogeneous entities from many sources. Matching and fusion of such entities require to also match and combine their properties (attributes). However, previous schema matching approaches mostly focus on two sources only and often rely on simple similarity measurements. They thus face problems in challenging use cases such as the integration of heterogeneous product entities from many sources.We therefore present a new machine learning-based property matching approach called LEAPME (LEArning-based Property Matching with Embeddings) that utilizes numerous features of both property names and instance values. The approach heavily makes use of word embeddings to better utilize the domain-specific semantics of both property names and instance values. The use of supervised machine learning helps exploit the predictive power of word embeddings.Our comparative evaluation against five baselines for several multi-source datasets with real-world data shows the high effectiveness of LEAPME. We also show that our approach is even effective when training data from another domain (transfer learning) is used.

show abstract

Ontology alignment using artificial neural network for large-scale ontologies

Cited by 19 publications

References 45 publications

Multi-level networked knowledge: representation and DL-reasoning

Multi-level networked knowledge: representation and DL-reasoning

Towards the smart use of embedding and instance features for property matching

LEAPME: Learning-based Property Matching with Embeddings

Contact Info

Product

Resources

About