Understood in translation: transformers for domain understanding

Dimitrios Christofidellis, Matteo Manica, Leonidas Georgopoulos, Hans Vandierendonck

Research output: Chapter in Book/Report/Conference proceedingConference contribution

44 Downloads (Pure)

Abstract

Knowledge acquisition is the essential first step of any Knowledge Graph (KG) application. This knowledge can be extracted from a given corpus (KG generation process) or specified from an existing KG (KG specification process). Focusing on domain specific solutions, knowledge acquisition is a labor intensive task usually orchestrated and supervised by subject matter experts. Specifically, the domain of interest is usually manually defined and then the needed generation or extraction tools are utilized to produce the KG. Herein, we propose a supervised machine learning method, based on Transformers, for domain definition of a corpus. We argue why such automated definition of the domain's structure is beneficial both in terms of construction time and quality of the generated graph. The proposed method is extensively validated on three public datasets (WebNLG, NYT and DocRED) by comparing it with two reference methods based on CNNs and RNNs models. The evaluation shows the efficiency of our model in this task. Focusing on scientific document understanding, we present a new health domain dataset based on publications extracted from PubMed and we successfully utilize our method on this. Lastly, we demonstrate how this work lays the foundation for fully automated and unsupervised KG generation.

Original languageEnglish
Title of host publicationProceedings of the Workshop on Scientific Document Understanding
Number of pages14
Publication statusPublished - 02 Feb 2021
EventWorkshop on Scientific Document Understanding, co-located with the 35th AAAI Conference on Artificial Intelligence - virtual, online
Duration: 09 Feb 202109 Feb 2021

Publication series

NameCEUR Workshop Proceedings
PublisherCEUR-WS
Volume2831
ISSN (Print)1613-0073

Conference

ConferenceWorkshop on Scientific Document Understanding, co-located with the 35th AAAI Conference on Artificial Intelligence
Abbreviated titleSDU 2021
Cityvirtual, online
Period09/02/202109/02/2021

ASJC Scopus subject areas

  • General Computer Science

Fingerprint

Dive into the research topics of 'Understood in translation: transformers for domain understanding'. Together they form a unique fingerprint.

Cite this