@article{fdi:010095821,
  title = {{A}nalysis of the performance of representation learning methods for entity alignment : benchmark versus real-world data},
  author = {{R}aoufi, {E}. and {H}appi {H}appi, {B}ill {G}ates and {L}armande, {P}ierre and {S}charffe, {F}. and {T}odorov, {K}.},
  editor = {},
  language = {{ENG}},
  abstract = {{R}epresentation learning for entity alignment ({EA}) aims to identify entities in different knowledge graphs ({KG}s) that refer to the same real-world object by comparing their embedding similarity. {A}lthough many {EA} models perform well on synthetic benchmark datasets, this performance does not always transfer to real-world, incomplete, and domain-specific data. {A} systematic comparison between synthetic benchmarks and original heterogeneous datasets is still limited. {M}any {EA} models also restrict the alignment search space to validation entities, limiting coverage of real {KG} content. {W}ithin this setting, our results show that embedding-based {EA} models continue to face generalization challenges in realistic large-scale {KG} search spaces. {W}e evaluate several competitive {EA} models-commonly tested on benchmarks such as {DBP}15{K}-on multiple real-world heterogeneous datasets. {T}he experiments reveal a performance decrease when moving beyond synthetic benchmarks, indicating that current models do not fully capture the characteristics of real data. {W}e also analyze semantic similarity and profiling features of the datasets to help explain these differences. {T}his study outlines practical limitations of embedding-based {EA} methods and provides insights for developing approaches that better handle the variability and complexity found in real-world {KG} alignment tasks.},
  keywords = {entity alignment ; knowledge graphs ; representation learning ; knowledge graph heterogeneity ; {EA} benchmarks},
  booktitle = {},
  journal = {{S}emantic {W}eb},
  volume = {17},
  numero = {1},
  pages = {09217134251389825 [24 p.]},
  ISSN = {1570-0844},
  year = {2026},
  DOI = {10.1177/09217134251389825},
  URL = {https://www.documentation.ird.fr/hor/fdi:010095821},
}