https://Mohammad-Abdul-Hadi.github.io/author/xuan-bach-d.-le/ Xuan-Bach D. Le Source Themes Academic (https://sourcethemes.com/academic/)en-us© Mohammad-Abdul-Hadi, 2026Sun, 01 May 2022 00:00:00 +0000 https://Mohammad-Abdul-Hadi.github.io/images/icon_hu_b453e4e1cf4dca05.png https://Mohammad-Abdul-Hadi.github.io/author/xuan-bach-d.-le/ https://Mohammad-Abdul-Hadi.github.io/publication/pretrained-models-api-learning/ Sun, 01 May 2022 00:00:00 +0000 https://Mohammad-Abdul-Hadi.github.io/publication/pretrained-models-api-learning/ <h2 id="overview">Overview</h2> <p>Pre-trained language models (PTMs) such as BERT, CodeBERT, and GPT variants have transformed NLP and are increasingly applied to software engineering tasks. This paper presents a systematic empirical study of PTM effectiveness specifically for <strong>API learning</strong> — the task of understanding, completing, and recommending API usage sequences from mixed natural-language and code inputs.</p> <h2 id="research-questions">Research Questions</h2> <ul> <li>How effective are PTMs at API sequence completion compared to non-PTM approaches?</li> <li>Does domain-specific pre-training (e.g., code-focused PTMs) outperform general PTMs for API learning?</li> <li>How well do PTMs generalize across programming languages for cross-lingual API mapping?</li> </ul> <h2 id="key-findings">Key Findings</h2> <ul> <li>PTMs consistently outperform traditional baselines on API learning tasks, particularly in low-resource settings.</li> <li>Code-specific PTMs (e.g., CodeBERT) provide measurable gains over general-purpose PTMs on code-centric subtasks.</li> <li>Cross-lingual transfer is effective, with PTMs showing strong generalization across Java and Python API benchmarks.</li> </ul> <p><strong>Published at:</strong> IEEE/ACM International Conference on Program Comprehension (ICPC) 2022 · <strong>Citations:</strong> 19</p>