Automated Detection of Self-Admitted Technical Debt: A Systematic Literature Review
الملخص
In modern software development, developers frequently introduce comments that describe temporary solutions, incomplete implementations, or deferred improvements. These comments, commonly referred to as Self-Admitted Technical Debt (SATD), may provide short-ter...
In modern software development, developers frequently introduce comments that describe temporary solutions, incomplete implementations, or deferred improvements. These comments, commonly referred to as Self-Admitted Technical Debt (SATD), may provide short-term flexibility but can accumulate into long-term maintenance challenges and quality degradation. This paper presents a systematic literature review of automated SATD detection research published between 2015 and 2025. Following established systematic review guidelines, we analyzed 46 primary studies selected from an initial pool of 1,537 publications. The review provides a structured synthesis of the field by:
- 1)identifying and categorizing the main SATD detection objectives;
- 2)analyzing representation techniques for textual, structural, and hybrid inputs;
- 3)comparing learning approaches ranging from traditional machine learning to transformer-based and large language models; and
- 4)summarizing evaluation practices and performance trends. Our findings reveal a clear methodological evolution from keyword-based and feature-engineered models toward context-aware and multi-modal architectures. While transformer-based approaches such as BERT and CodeBERT demonstrate strong potential, their adoption is constrained by limited benchmark datasets, inconsistent evaluation protocols, and a lack of explainability. To address these challenges, the review highlights key research gaps and outlines future directions, including standardized benchmarks, explainable SATD detection, and cross-project generalization. Overall, this study consolidates a decade of SATD research into a unified perspective and provides actionable guidance for developing more robust, interpretable, and scalable SATD detection techniques to support sustainable software maintenance.
الكلمات المفتاحية
الباحثون
- أنس عبد العزيز المؤلف الرئيسي
- رياض سنبل مؤلف مشارك
- مهيب النقري مؤلف مشارك
ملفات المنشور
منشورات ذات صلة
Wathiq-SPU at AraSeg Shared Task 2026: Calibrated Transformer Ensembles for Arabic Sentence Segmentation across Punctuation and Paragraph Conditions
2026
WathiqTeam at AraGenre 2026: Class-Balanced Hybrid Retrieval with Selective LLM Reranking
2026
SPU_DeepSimplify at the CLEF 2026 SimpleText Track: A Planning-Based Approach to Scientific Text Simplification
2026