ViewEntanglement: Latent-Space Alignment and Supervised Re-ranking for Biomedical Ontology Matching

Ontology matching remains a fundamental challenge in biomedical data integration, since independently developed ontologies, such as the Mouse Anatomy (MA) and the National Cancer Institute Thesaurus (NCI), adopt different terminologies, granulari-ties and conceptual structures. This work presents ViewEntanglement, a revised version of the View_Entanglement_7 pipeline, which combines sentence-level em-beddings (SBERT), explicit latent-space entanglement simulated through random orthogonal transformations, a ridge-regression codec to align the entangled spaces, and a supervised re-ranker based on Random Forest. We formalize the orthogona-lity property of the true cross-referential transformation, analyze the computational complexity of each stage, and evaluate the pipeline on the Anatomy benchmark. Reference results indicate that the learned codec almost fully reverses the simulated entanglement (relative Frobenius error ≈ 0,12) and that the Random Forest re-ranker raises the F1-Score by approximately 3 percentage points over the cosine baseline.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-18
DOI
https://doi.org/10.5281/zenodo.24330510
Primary Topic
Biomedical Text Mining and Ontologies
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

ViewEntanglement: Latent-Space Alignment and Supervised Re-ranking for Biomedical Ontology Matching

Roberto Monteiro Dias
Zenodo (CERN European Organization for Nuclear Research)
Biomedical Text Mining and Ontologies
preprint

ViewEntanglement: Latent-Space Alignment and Supervised Re-ranking for Biomedical Ontology Matching

Roberto Monteiro Dias
preprint en

Abstract

Ontology matching remains a fundamental challenge in biomedical data integration, since independently developed ontologies, such as the Mouse Anatomy (MA) and the National Cancer Institute Thesaurus (NCI), adopt different terminologies, granulari-ties and conceptual structures. This work presents ViewEntanglement, a revised version of the View_Entanglement_7 pipeline, which combines sentence-level em-beddings (SBERT), explicit latent-space entanglement simulated through random orthogonal transformations, a ridge-regression codec to align the entangled spaces, and a supervised re-ranker based on Random Forest. We formalize the orthogona-lity property of the true cross-referential transformation, analyze the computational complexity of each stage, and evaluate the pipeline on the Anatomy benchmark. Reference results indicate that the learned codec almost fully reverses the simulated entanglement (relative Frobenius error ≈ 0,12) and that the Random Forest re-ranker raises the F1-Score by approximately 3 percentage points over the cosine baseline.

Zenodo (CERN European Organization for Nuclear Research)
Universidade Federal do Estado do Rio de Janeiro (BR)
Biomedical Text Mining and Ontologies
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon