Automatic Detection and Generation of Argument Structures Within the Medical Domain

Walter Koza, Constanza Suy

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Representing the predicate argument structure of the medical domain (ASMD) is important for automatic text analyses. This work aims to describe the ASMD through verbs and transformational possibilities. Computer resources were constructed from 100 selected biomedical verbs (corpus CCM2009). Firstly, these were analyzed to determine the quantity and type of arguments, and classified these arguments into object classes (OC). Secondly, we established possible transformations for each ASMD. With this information, we created computer models on NooJ for the detection and automatic creation of ASMD in a corpus. This work involved the elaboration of electronic dictionaries, syntactic recognition, and generative grammars. The detection was performed on a corpus of 188,000 words conformed by texts from the gynecology and obstetrics area, achieving the following results: 100% accuracy, 96.92% coverage, and 98% F-measure. NooJ grammars provided grammatical sentences of each ASMD involving different transformations admitted by each particular class.

Original languageEnglish
Title of host publicationFormalizing Natural Languages
Subtitle of host publicationApplications to Natural Language Processing and Digital Humanities - 15th International Conference, NooJ 2021, Revised Selected Papers
EditorsMagali Bigey, Annabel Richeton, Max Silberztein, Izabella Thomas
PublisherSpringer Science and Business Media Deutschland GmbH
Pages198-207
Number of pages10
ISBN (Print)9783030928605
DOIs
StatePublished - 2021
Externally publishedYes
Event15th International Conference, NooJ 2021 - Virtual Online
Duration: 9 Jun 202111 Jun 2021

Publication series

NameCommunications in Computer and Information Science
Volume1520 CCIS
ISSN (Print)1865-0929
ISSN (Electronic)1865-0937

Conference

Conference15th International Conference, NooJ 2021
CityVirtual Online
Period9/06/2111/06/21

Keywords

  • Biomedical domain
  • Lexicon-grammar
  • NooJ
  • Predicate argument structure

Fingerprint

Dive into the research topics of 'Automatic Detection and Generation of Argument Structures Within the Medical Domain'. Together they form a unique fingerprint.

Cite this