Ebube Alor

Ebube Alor

Auto-generated profile, based on publicly available information

Ebube Alor

Concordia University, Canada

Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
6 June 2026 in ACM Transactions on Software Engineering and Methodology
Verified
Journal Article
Open access
EAEbube Alor
Ebube Alor
SKSayedHassan Khatoonabadi
SayedHassan Khatoonabadi
ESEmad Shihab
Emad Shihab
Large Language Models (LLMs) offer new potential for automating documentation-to-code traceability, yet their capabilities and limitations remain underexplored. We evaluate three LLMs (Claude Sonnet, GPT-4o, and o3-mini) on documentation-to-code traceability across diverse software projects. We create two datasets from open-source systems (Unity Catalog and Crawl4AI), curated to minimize training-data overlap with the evaluated models, and derive a 27-project benchmark from CoDocBench and TransArc to study generalizability across established codebases. Our experiments assess three capabilities: (1) trace link identification, (2) relationship explanation quality, and (3) multi-step chain reconstruction, while comparing LLMs against TF-IDF, BM25, and CodeBERT baselines. On the curated datasets, the best LLM attains F1-scores of 79.4% and 80.4%, and on the Established Projects benchmark, LLMs reach average F1 up to 92.3%, substantially outperforming traditional baselines except in a single dense system. Fully correct explanations cover 42.9%–71.1% of links, while partial correctness exceeds 97%, and chain endpoints are almost always recovered, though intermediate steps are often imperfect. We also show that task framing, particularly one-to-many matching, is crucial for recall, whereas naive many-to-many prompts achieve around 5% F1 due to enumeration bias. Overall, LLMs are promising assistants for trace discovery when combined with human oversight and careful tool design.

12 April 2026
Verified
Conference Paper
Open access
EAEbube Alor
Ebube Alor
JTJoão Pedro de Souza Olivo Tardivo
João Pedro de Souza Olivo Tardivo
SKSayedHassan Khatoonabadi
SayedHassan Khatoonabadi
ESEmad Shihab
Emad Shihab
When developers change source code, related artifacts like documentation and tests often become outdated, creating inconsistencies that are difficult to fix manually. We introduce ArtifactSync, a tool that leverages Large Language Models (LLMs) for automated cross-artifact consistency maintenance. LLMs are uniquely suited for this task because they can understand semantic relationships across diverse artifact types, such as code, documentation, and tests that traditional static analysis cannot capture. ArtifactSync performs change impact analysis using a hierarchical approach that starts by scanning file names and progressively “zooms in” on uncertain files, requesting more detailed context only when necessary. This targeted change impact analysis makes the evaluation of large repositories both feasible and cost-effective. ArtifactSync is implemented as both a command-line tool and an interactive VS Code extension. Within the editor, it presents each detected inconsistency with a clear explanation and an actionable fix that developers can apply directly. Our preliminary evaluation across two open-source repositories shows that ArtifactSync achieved up to 100% accuracy in both identifying impacted artifacts and generating corresponding fixes. A demonstration is available at https://youtu.be/AfLe_I_OnA0 and the tool at https://github.com/Alor-e/artifact-sync.