Context-Aware Classification Method for Survey Data Using Natural Language Embeddings
Journal of Data Forensics Research · 31.12.2024 · 1 (1) · 109-123
DOI: 10.12972/jdfr.2024.1.1.7 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Data Forensics Research, online erschienen am 31.12.2024. Verfasst von Jaekyeong Jung und 3 weiteren Autor:innen.
Abstract (Originalsprache)
Survey response data is typically quantified for analysis; however, when each question is represented as a single score, the context reflecting the intent and importance may be overlooked. To address this, we propose a method that generates natural language embeddings by concatenating questions and responses through large language models (LLMs). By comparing embedding vectors derived from question-response pairs using various pre-trained LLMs, we identify the model most suitable for capturing contextual information. Applying the proposed approach to mental health survey data, we identified the collective characteristics of individuals with suicidal thoughts, plans, and attempts, demonstrating that embedding vectors extracted from question-response pairs are more effective in classifying mental health risk groups compared to traditional score-based evaluation methods.
Zitieren
APA
Jung, J., Oh, M., Shin, S., & Choi, H. (2024). Context-Aware Classification Method for Survey Data Using Natural Language Embeddings. Journal of Data Forensics Research, 1(1), 109-123. https://doi.org/10.12972/jdfr.2024.1.1.7
Vancouver
Jung J, Oh M, Shin S, Choi H. Context-Aware Classification Method for Survey Data Using Natural Language Embeddings. Journal of Data Forensics Research. 2024;1(1):109-123. doi:10.12972/jdfr.2024.1.1.7
BibTeX
@article{Jung2024,
author = {Jung, Jaekyeong and Oh, Miae and Shin, Sungcheol and Choi, Hosik},
title = {Context-Aware Classification Method for Survey Data Using Natural Language Embeddings},
journal = {Journal of Data Forensics Research},
year = {2024},
volume = {1},
number = {1},
pages = {109--123},
doi = {10.12972/jdfr.2024.1.1.7}
}
EUVIMED