From Heuristics to Language Models: A Journey Through the Universe of Semantic Table Interpretation with DAGOBAH - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

From Heuristics to Language Models: A Journey Through the Universe of Semantic Table Interpretation with DAGOBAH

Viet-Phi Huynh
  • Fonction : Auteur
  • PersonId : 1272828
Yoan Chabot
  • Fonction : Auteur
Thomas Labbé
  • Fonction : Auteur
Jixiong Liu
  • Fonction : Auteur

Résumé

This paper presents DAGOBAH SL 2022, a semantic table interpretation system that has been continuously improved over the last four years when participating in the SemTab challenge. This year, we have improved the lookup coverage using external resources and we have integrated language models for better understanding the table headers. We have also implemented various system optimizations that lead to a reduction in execution time of about 30%. In this paper, we also show the relevance of using deep learning-based approaches for resolving certain ambiguities and we discuss the limitations of existing corpora and systems for maturing further this research field.
Fichier principal
Vignette du fichier
iswc2022.pdf (1.93 Mo) Télécharger le fichier
Origine : Fichiers éditeurs autorisés sur une archive ouverte

Dates et versions

hal-04170873 , version 1 (25-07-2023)

Identifiants

  • HAL Id : hal-04170873 , version 1

Citer

Viet-Phi Huynh, Yoan Chabot, Thomas Labbé, Jixiong Liu, Raphaël Troncy. From Heuristics to Language Models: A Journey Through the Universe of Semantic Table Interpretation with DAGOBAH. 21st International Semantic Web Conference (ISWC 2022), Oct 2022, Hangzhou (virtual), China. ⟨hal-04170873⟩

Collections

EURECOM
20 Consultations
16 Téléchargements

Partager

Gmail Facebook X LinkedIn More