Replication Package: Comparing TF-IDF, Sentence-BERT, and LSTM for Self-Admitted Technical Debt Classification

Replication Package This replication package contains supplementary materials for the study "Comparing TF-IDF, Sentence-BERT, and LSTM for Self-Admitted Technical Debt Classification". Dataset The dataset is provided as: technical_debt_dataset.csv Columns: projectname classification commenttext Experiments Computational cost metrics were collected in a Google Colab environment, version 2026.04, running on Ubuntu 22.04.5 LTS, using Python 3.12.13, NumPy 2.0.2, PyTorch 2.10.0, JAX 0.7.2, and TensorFlow 2.19.0. R 4.5.3, codename Reassured Reassurer and Julia 1.11.5 were also available. Experiment 01:Comparison of textual representations and classifiers. Experiment 02:Impact of data balancing using ADASYN. Requirements Python 3.11+ Install dependencies: pip install -r requirements.txt Reproducibility Each experiment can be executed independently using the provided scripts. All reported metrics were generated using the files included in this package.