Extraction of Aspect Categories for Sentiment Analysis towards Technical and Vocational Education and Training (TVET) in Malaysia using Topic Modelling Approach

  • Nurul Ashikin Abd Samad Politeknik Mersing
  • Roliana Ibrahim Universiti Teknologi Malaysia

Abstract

This study highlights the neccesity of the analyses on public opinion through more transparent approach in order to observe the public perception towards technical and vocational education and training (TVET) in Malaysia. Aspect-based sentiment analysis (ABSA) enables the identification the aspect or features that reflect positive and negative sentiments in sentences. However, one of the challenges in ABSA is the extraction of aspect from the dataset. In this study, topic modelling is used to identify the aspect words from tweets related to TVET in Malaysia. One of the topic model that is proven useful in  extracting topics from corpus of lesser known domain is Latent Dirichlet Allocation (LDA). LDA extracted aspects terms were compared with manually labeled aspect terms. The study shows means of accuracy at 0.96 while the means of precision is acceptable at 0.70 respectively after applying LDA in extracting the aspec categories. It is hope that by harvesting opinions from social media will enable the discovery of the most frequent aspects about TVET in Malaysia that often mentioned by the public. This research is hoping to contribute towards the improvisation of Malaysia’s TVET in the future.

Published
2020-12-01
How to Cite
ABD SAMAD, Nurul Ashikin; IBRAHIM, Roliana. Extraction of Aspect Categories for Sentiment Analysis towards Technical and Vocational Education and Training (TVET) in Malaysia using Topic Modelling Approach. Politeknik & Kolej Komuniti Journal of Engineering and Technology, [S.l.], v. 5, n. 1, p. 119-129, dec. 2020. ISSN 0128-2883. Available at: <https://myjms.mohe.gov.my/index.php/PMJET/article/view/11354>. Date accessed: 24 oct. 2021.

Most read articles by the same author(s)

Obs.: This plugin requires at least one statistics/report plugin to be enabled. If your statistics plugins provide more than one metric then please also select a main metric on the admin's site settings page and/or on the journal manager's settings pages.