Bringing Back Semantics to Knowledge Graph Embeddings: An Interpretability Approach

Antoine Domingues; Nitisha Jain; Albert Merono Penuela; Elena Simperl

doi:10.1007/978-3-031-71170-1_17

Bringing Back Semantics to Knowledge Graph Embeddings: An Interpretability Approach

Antoine Domingues^*, Nitisha Jain^*, Albert Merono Penuela, Elena Simperl

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › peer-review

162 Downloads (Pure)

Abstract

Knowledge Graph Embeddings Models project entities and relations from Knowledge Graphs into a vector space. Despite their widespread application, concerns persist about the ability of these models to capture entity similarity effectively. To address this, we introduce InterpretE, a novel neuro-symbolic approach to derive interpretable vector spaces with human-understandable dimensions in terms of the features of the entities. We demonstrate the efficacy of InterpretE in encapsulating desired semantic features, presenting evaluations both in the vector space as well as in terms of semantic similarity measurements.

Original language	English
Title of host publication	18th International Conference on Neural-Symbolic Learning and Reasoning (NeSy 2024)
Publisher	Springer, Cham
Pages	192-203
Number of pages	12
Volume	14979
ISBN (Electronic)	978-3-031-71167-1
ISBN (Print)	978-3-031-71166-4
DOIs	https://doi.org/10.1007/978-3-031-71170-1_17 https://doi.org/10.1007/978-3-031-71170-1_17
Publication status	Published - 10 Sept 2024
Event	18th International Conference on Neural-Symbolic Learning and Reasoning, NeSy 2024 - Barcelona, Spain Duration: 9 Sept 2024 → 12 Sept 2024

Publication series

Name	Neural-Symbolic Learning and Reasoning
Volume	14980
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	18th International Conference on Neural-Symbolic Learning and Reasoning, NeSy 2024
Country/Territory	Spain
City	Barcelona
Period	9/09/2024 → 12/09/2024

Keywords

interpretable vectors
knowledge graph embeddings
semantic similarity

Access to Document

NeSy_submissionAccepted author manuscript, 1.83 MB
637244_1_En_17_Chapter_OnlinePDFProof, 1.44 MB

https://link.springer.com/10.1007/978-3-031-71170-1_17

Cite this

Domingues, A., Jain, N., Merono Penuela, A., & Simperl, E. (2024). Bringing Back Semantics to Knowledge Graph Embeddings: An Interpretability Approach. In 18th International Conference on Neural-Symbolic Learning and Reasoning (NeSy 2024) (Vol. 14979, pp. 192-203). Article Chapter 17 (Neural-Symbolic Learning and Reasoning; Vol. 14980). Springer, Cham. https://doi.org/10.1007/978-3-031-71170-1_17, https://doi.org/10.1007/978-3-031-71170-1_17

@inbook{755db6c66d7543b6b986e4dcb2c875d1,

title = "Bringing Back Semantics to Knowledge Graph Embeddings: An Interpretability Approach",

abstract = "Knowledge Graph Embeddings Models project entities and relations from Knowledge Graphs into a vector space. Despite their widespread application, concerns persist about the ability of these models to capture entity similarity effectively. To address this, we introduce InterpretE, a novel neuro-symbolic approach to derive interpretable vector spaces with human-understandable dimensions in terms of the features of the entities. We demonstrate the efficacy of InterpretE in encapsulating desired semantic features, presenting evaluations both in the vector space as well as in terms of semantic similarity measurements.",

keywords = "interpretable vectors, knowledge graph embeddings, semantic similarity",

author = "Antoine Domingues and Nitisha Jain and {Merono Penuela}, Albert and Elena Simperl",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.; 18th International Conference on Neural-Symbolic Learning and Reasoning, NeSy 2024 ; Conference date: 09-09-2024 Through 12-09-2024",

year = "2024",

month = sep,

day = "10",

doi = "10.1007/978-3-031-71170-1_17",

language = "English",

isbn = "978-3-031-71166-4",

volume = "14979",

series = "Neural-Symbolic Learning and Reasoning",

publisher = "Springer, Cham",

pages = "192--203",

booktitle = "18th International Conference on Neural-Symbolic Learning and Reasoning (NeSy 2024)",

}

Domingues, A, Jain, N , Merono Penuela, A & Simperl, E 2024, Bringing Back Semantics to Knowledge Graph Embeddings: An Interpretability Approach. in 18th International Conference on Neural-Symbolic Learning and Reasoning (NeSy 2024). vol. 14979, Chapter 17, Neural-Symbolic Learning and Reasoning, vol. 14980, Springer, Cham, pp. 192-203, 18th International Conference on Neural-Symbolic Learning and Reasoning, NeSy 2024, Barcelona, Spain, 9/09/2024. https://doi.org/10.1007/978-3-031-71170-1_17, https://doi.org/10.1007/978-3-031-71170-1_17

Bringing Back Semantics to Knowledge Graph Embeddings: An Interpretability Approach. / Domingues, Antoine; Jain, Nitisha ; Merono Penuela, Albert et al.
18th International Conference on Neural-Symbolic Learning and Reasoning (NeSy 2024). Vol. 14979 Springer, Cham, 2024. p. 192-203 Chapter 17 (Neural-Symbolic Learning and Reasoning; Vol. 14980).

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › peer-review

TY - CHAP

T1 - Bringing Back Semantics to Knowledge Graph Embeddings

T2 - 18th International Conference on Neural-Symbolic Learning and Reasoning, NeSy 2024

AU - Domingues, Antoine

AU - Jain, Nitisha

AU - Merono Penuela, Albert

AU - Simperl, Elena

PY - 2024/9/10

Y1 - 2024/9/10

N2 - Knowledge Graph Embeddings Models project entities and relations from Knowledge Graphs into a vector space. Despite their widespread application, concerns persist about the ability of these models to capture entity similarity effectively. To address this, we introduce InterpretE, a novel neuro-symbolic approach to derive interpretable vector spaces with human-understandable dimensions in terms of the features of the entities. We demonstrate the efficacy of InterpretE in encapsulating desired semantic features, presenting evaluations both in the vector space as well as in terms of semantic similarity measurements.

AB - Knowledge Graph Embeddings Models project entities and relations from Knowledge Graphs into a vector space. Despite their widespread application, concerns persist about the ability of these models to capture entity similarity effectively. To address this, we introduce InterpretE, a novel neuro-symbolic approach to derive interpretable vector spaces with human-understandable dimensions in terms of the features of the entities. We demonstrate the efficacy of InterpretE in encapsulating desired semantic features, presenting evaluations both in the vector space as well as in terms of semantic similarity measurements.

KW - interpretable vectors

KW - knowledge graph embeddings

KW - semantic similarity

UR - http://www.scopus.com/inward/record.url?scp=85204904471&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-71170-1_17

DO - 10.1007/978-3-031-71170-1_17

M3 - Conference paper

SN - 978-3-031-71166-4

VL - 14979

T3 - Neural-Symbolic Learning and Reasoning

SP - 192

EP - 203

BT - 18th International Conference on Neural-Symbolic Learning and Reasoning (NeSy 2024)

PB - Springer, Cham

Y2 - 9 September 2024 through 12 September 2024

ER -

Domingues A, Jain N , Merono Penuela A , Simperl E. Bringing Back Semantics to Knowledge Graph Embeddings: An Interpretability Approach. In 18th International Conference on Neural-Symbolic Learning and Reasoning (NeSy 2024). Vol. 14979. Springer, Cham. 2024. p. 192-203. Chapter 17. (Neural-Symbolic Learning and Reasoning). doi: 10.1007/978-3-031-71170-1_17, 10.1007/978-3-031-71170-1_17