본문 바로가기
  • Home

A Study on the Correlation between Meteorological Factors and Sentiment using KoBERT-based Multidimensional Sentiment Analysis

  • Journal of The Korea Society of Computer and Information
  • Abbr : JKSCI
  • 2026, 31(4), pp.183~189
  • Publisher : The Korean Society Of Computer And Information
  • Research Area : Engineering > Computer Science
  • Received : February 12, 2026
  • Accepted : April 11, 2026
  • Published : April 30, 2026

Hyun Jin Yeo 1

1배재대학교

Accredited

ABSTRACT

This study analyzes SNS text data, categorized by date and region, integrated with corresponding meteorological data, to investigate changes in human sentiment driven by weather factors from a data science perspective. While prior research on this topic exists, most studies have relied on lexicon-based approaches, classifying weather-induced emotions merely into binary positive or negative categories. In contrast, this study aims to analyze a diverse range of emotions associated with weather conditions by classifying sentiments into multiple categories using Natural Language Processing (NLP) based on BERT (Bidirectional Encoder Representations from Transformers). To achieve this, we utilized the 'Sentiment Dialogue Corpus' from AI Hub and a dataset classified into seven distinct emotions using KoBERT. The analysis demonstrated significant model performance based on meteorological variables such as temperature, precipitation, and humidity.

Citation status

* References for papers published after 2024 are currently being built.