HOME
Teaching Research Publications CV
Album
Software Project Contact me Find me on social network

RESEARCH

In this page, you find details about my research works conducted during my doctoral studies, Post-Doctoral Research (HDR preparation) and current work. To foster transparency, reproducibility, and collaboration within the research community, I consistently make publically available all datasets, scripts, and tools utilized in my research. By sharing these resources, I aim to facilitate the replication of my findings, encourage further investigation, and stimulate innovative advancements in the field.

Doctoral studies

My doctoral dissertation was titled 'A Contribution to Software Vulnerability Prediction: A Code Metrics-Based Approach.' Within the scope of this thesis, I proposed approaches that leverage software metrics and ML/DL techniques for automatic prediction of software vulnerabilities. Automatic vulnerability prediction can significantly assist developers and minimize the resources allocated to addressing software security issues. These costs can be further reduced by accurately pinpointing the exact locations of vulnerabilities (vulnerable lines of code).

A key strength of the proposed approaches lies in their ability to utilize code metrics to quantify code slices that suggest the presence of vulnerabilities at a fine-grained level (a few lines of code). The work conducted in my thesis has resulted in the following publications:

  1. Paper 1 :
    • Title : Evaluating and Comparing Size, Complexity and Coupling Metrics as Web Applications Vulnerabilities Predictors
    • Journal : International Journal of Information Technology and Computer Science(IJITCS)
    • Authors : Mohammed Zagane , Mustapha Kamel Abdi
    • URL : https://www.mecs-press.org/ijitcs/ijitcs-v11-n7/IJITCS-V11-N7-5.pdf
    • DOI : 10.5815/ijitcs.2019.07.05

  2. Paper 2 :
  3. Paper 3 :
  4. Paper 4 :
    • Title : Évaluation des Metriques de Couplage en Tant Qu’indicateurs de Vulnérabilités dans les Applications Web
    • Journal : Communication Science et Technologie (COST)
    • Authors : Mohammed Zagane and Mustapha Kamel Abdi
    • DOI/URL : https://asjp.cerist.dz/en/article/150783
  5. Tools :
  6. Datasets:


Post-Doctoral research (HDR Preparation)

Following my doctoral studies, I sought to address the limitations of software metric-based approaches, which are specific to the domain of software engineering. To this end, I drew inspiration from the field of Natural Language Processing (NLP), where traditional and deep learning techniques have yielded impressive results. This inspiration was motivated by the similarities between software source code and natural language: both exhibit syntactic and semantic characteristics, as well as a defined vocabulary. In collaboration with Saudi researchers, I proposed a novel approach [Paper 5] that leverages Word Embeddings, a widely used technique in NLP, for vulnerability prediction. Moreover, in the primary article of my HDR [Paper 6], I introduced another approach inspired by NLP that employs Term Frequency-Inverse Document Frequency (TF-IDF), a common technique in both NLP and information retrieval. This approach enables the automatic extraction of relevant attributes and the construction of effective vulnerability prediction models. To demonstrate the efficacy of this proposed approach, I conducted comparative studies with traditional software metrics and found that the automatically extracted attributes significantly outperformed these metrics."

These works have led to the following publications and Datasets:

  1. Paper 5 :
    • Title : Efficient Deep Features Learning for Vulnerability Detection Using Character N-Gram Embedding
    • Journal : Jordanian Journal of Computers and Information Technology (JJCIT)
    • Authors : Mamdouh Alenezi and Mohammed Zagane
      • URL : https://www.jjcit.org/paper/120/EFFICIENT-DEEP-FEATURES-...
    • DOI : 10.5455/jjcit.71-1597824949

  2. Paper 6 :
  3. Conference Paper 1 :
    • Title : Automatic Feature Extraction Method for Software Vulnerability Prediction
    • Conference : 1st National Conference on Applied Computing and Smart Technologies, SBA, Algeria
    • Authors : Mohammed Zagane , Mustapha Kamel Abdi and Mamdouh Alenezi
  4. Conference Paper 2 :
    • Title : Apprentissage Profond pour la Détection du Code Source Vulnérable : Approche inspirée du domaine de TLN
    • Conference : Mathématiques et Informatique Appliquée aux Sciences, MIAS’2022, Université de Tamanghasset, Algeria
    • Authors : Mohammed Zagane and AbdErrahim Zagane
  5. Datasets and Tools:

Current work (Post-HDR research)

In addition to my continuing research in automated vulnerability prediction (AVP), I have diversified my research interests to encompass the application of machine learning (ML) and deep learning (DL) techniques to tackle challenging problems within other software engineering subfields. Notably, I have conducted research in software change management, where I have introduced a novel hybrid approach that leverages both software metrics and word embeddings to significantly improve the performance of co-change prediction models. This work is detailed in [paper 7].

  1. Paper 7 :

Furthermore, I have played a pivotal role in establishing the LABTEC-IA research laboratory and serve as the head of the AIIS team. As I move forward, my primary objective is to mentor PhD students and guide their research in exploring the potential of emerging AI technologies to address critical societal challenges such as cybersecurity and food security."