Intent Mining of Thai Phone Call Text Using a Stacking Ensemble Classifier with GPT-3 Embeddings

Nattapong Sanchan

Abstract

Intent mining has recently attracted Natural Language Processing (NLP) research communities. Despite extensive research on English and other widely spoken languages, intent mining in Thai remains unexplored. This paper proposes an extended framework for mining intentions in Thai phone-call text. It utilizes a stacking ensemble method with GPT-3 embeddings, constructed by systematically determining base and meta classifiers using Q-statistic and F1 scores. The base classifiers — Support Vector Classifier (SVC), k-Nearest Neighbors (KNN), and Random Forest (RF) — were combined with a meta-classifier, Logistic Regression (LR). We compared the mining results, derived through the proposed Stacking Ensemble Classifier (SEC), to (1) the individual base classifiers and (2) three BERT baselines: BERT Multilingual Uncased, BERT-th, and BERT Base EN-TH Cased. The results revealed that SEC outperformed SVC, KNN, RF, BERT Multilingual Uncased, and BERT-th, though not BERT Base EN-TH Cased. A statistical analysis using Friedman and Holm's post-hoc tests found no statistically significant difference between SEC and BERT Base EN-TH Cased, suggesting the two classifiers perform similarly in practice.

Keywords: intent mining, intention mining, intent classification, intent detection, text mining

References

Download paper (PDF) Download BibTeX

BibTeX entry

@article{2025_Sanchan,
  title   = {Intent Mining of Thai Phone Call Text Using a Stacking Ensemble Classifier with GPT-3 Embeddings},
  author  = {Sanchan, Nattapong},
  journal = {ECTI Transactions on Computer and Information Technology (ECTI-CIT)},
  volume  = {19},
  number  = {1},
  pages   = {135--145},
  year    = {2025},
  month   = {Jan},
  doi     = {10.37936/ecti-cit.2025191.258239}
}

Rich-text citation (copy & paste)

Sanchan, N. (2025). Intent Mining of Thai Phone Call Text Using a Stacking Ensemble Classifier with GPT-3 Embeddings. ECTI Transactions on Computer and Information Technology (ECTI-CIT), 19(1), 135–145. https://doi.org/10.37936/ecti-cit.2025191.258239

More information

This paper is a continuing work in the Thai Intention Mining project: