본문 바로가기
  • Home

Research on Constructing a Sentiment Lexicon for the F&B Sector based on the N-gram Framework

  • Journal of The Korea Society of Computer and Information
  • Abbr : JKSCI
  • 2024, 29(10), pp.11-19
  • DOI : 10.9708/jksci.2024.29.10.011
  • Publisher : The Korean Society Of Computer And Information
  • Research Area : Engineering > Computer Science
  • Received : July 31, 2024
  • Accepted : September 25, 2024
  • Published : October 31, 2024

Yeryung Moon 1 Gaeun Son 1 Geonuk Nam 1 LEE, HAN JIN 1

1한동대학교

Accredited

ABSTRACT

Online and mobile reviews strongly influence consumer behavior, especially in the service industry, and play a key role in determining customer retention and revisit rates. Systematically analyzing the information in these reviews can effectively assess how they directly influence customers' purchase decisions. In this study, we applied the existing KNU sentiment dictionary to food and beverage (F&B) review data to build a customized sentiment lexicon using N-grams based on about 10,000 reviews. Comparing its performance with the existing dictionary, we found that the sentiment lexicon generated using the 1-gram, 2-gram, and 3-gram models had the highest accuracy, precision, recall, and F1 scores. These results can serve as a powerful business support tool for SMEs in the F&B and grocery shopping sector, also be used to predict customer demand for technology and policy.

Citation status

* References for papers published after 2023 are currently being built.