Semantic Role Labeling for Information Extraction on Indonesian Texts: A Literature Review

Amelia Devi Putri Ariyanto, Chastine Fatichah, Diana Purwitasari

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

4 Citations (Scopus)

Abstract

The information extraction process includes Semantic Role Labeling (SRL) as one of its sub-tasks. SRL aims to determine the semantic role of each entity within a sentence by examining the meaning of the predicate. This helps construct the sentence structure by identifying the relationships between predicates and their corresponding arguments. SRL development is less common than Named Entity Recognition (NER) for information extraction because SRL annotation process is complicated, and labeling results are sometimes ambiguous. In event extraction problem, the use of NER alone is insufficient. Identifying location entities generated by NER is still inaccurate because geographic coordinates indicate locations irrelevant to actual events. On the other hand, SRL can detect locations precisely and in depth according to actual events. Even though the annotation process is complicated, the SRL can be adjusted according to the required domain and its ontology so that SRL can extract location entities down to the event level.. This research aims to offer a comprehensive analysis concerning the advancement of Semantic Role Labeling (SRL) for extracting information from Indonesian texts. Indonesian is a low-resource language with a different character from English and only has very little literature, so it is interesting to study. The papers used for the review process came from IEEE, Science Direct, and Google Scholar from 2013 to 2023, and 15 papers were found that matched the research objectives. The study results show that most papers use Indonesian-language news articles as their dataset because they use formal language, which usually has a good language structure. The methods used in SRLs are mostly rule-based. A weakness of the rule-based development method is that the rules are very dependent on a particular language or problem domain. Thus, further work can use a transformer-based deep learning approach to perform SRL on Indonesian-language texts.

Original languageEnglish
Title of host publication2023 International Seminar on Intelligent Technology and Its Applications
Subtitle of host publicationLeveraging Intelligent Systems to Achieve Sustainable Development Goals, ISITIA 2023 - Proceeding
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages119-124
Number of pages6
ISBN (Electronic)9798350313956
DOIs
Publication statusPublished - 2023
Event24th International Seminar on Intelligent Technology and Its Applications, ISITIA 2023 - Hybrid, Surabaya, Indonesia
Duration: 26 Jul 202327 Jul 2023

Publication series

Name2023 International Seminar on Intelligent Technology and Its Applications: Leveraging Intelligent Systems to Achieve Sustainable Development Goals, ISITIA 2023 - Proceeding

Conference

Conference24th International Seminar on Intelligent Technology and Its Applications, ISITIA 2023
Country/TerritoryIndonesia
CityHybrid, Surabaya
Period26/07/2327/07/23

Keywords

  • Indonesian Text
  • Information Extraction
  • Literature Review
  • Named Entity Recognition
  • Semantic Role Labeling

Fingerprint

Dive into the research topics of 'Semantic Role Labeling for Information Extraction on Indonesian Texts: A Literature Review'. Together they form a unique fingerprint.

Cite this