HEMZA Loucif

ABOUBAKER BENAMRA

The conception of a monitoring system based on biometric recognition using a Hybrid approach

aziz salah eddine , tharrafi abdellah

developement of an online store for product and order management

Elbar Zakaria , Mabkhout et Khier Yacine

Cryptool Account Manager

Ben elbar Mohammed

موقع ويب للتصويت مبني باستخدام البلوك شين

Bennaama Mohamed

The conception of a monitoring system based on biometric recognition using a Hybrid approach

Ahmed Fetayah

Lane line detection system

OUNICI Khaled

Towards a new simplistic approach for influential web users identification in social networks

Fali Bilal

Developing an Intelligent Application for Human Being Identification using a Hybrid Approach

Zebbiche Lydia

The analysis of information diffusion in social media networks: A comparative experimental study

Spammers in online social networks continuously adapt their strategies, making detection a challenging and dynamic task. While traditional machine learning models and static deep learning approaches such as CNNs achieve good performance, they often fail to capture the temporal evolution of user behavior and network interactions. In this paper, we propose a novel deep learning framework for dynamic spammer detection that combines Principal Component Analysis (PCA) for feature reduction, Convolutional Neural Networks (CNNs) for local content feature extraction, and Temporal Graph Embeddings (TGEs) to capture evolving interaction patterns over time. Unlike prior static models, our approach explicitly models the dynamics of user behavior and relational changes in the social graph. Experiments conducted on benchmark Twitter datasets demonstrate that our hybrid PCA CNN–TGE model significantly outperforms classical baselines (ANN, CNN, SVM) and static hybrid models, achieving an F1-score of 94 %. The results highlight the importance of temporal graph learning for robust and adaptive spammer detection in social networks.

Citation

Hemza Loucif , Samir Akhrouf , ,(2025-11-25), Dynamic spammer detection using deep learning with temporal graph embeddings,Second International Workshop on Machine Learning and Deep Learning,Mohamed Boudif University M'Sila, Faculté MI

The dramatic growth of social networks and the diversity of their user base give the web business companies an unprecedented opportunity to stay well ahead of competition. Companies like Amazon, Alibaba, and other web business leaders are still investing huge amounts of money in order to improve their social network-based online recommender systems. Detecting and filtering spammers (aka fake recommenders) who post messages containing malicious commercial URLs in these environments is becoming a serious issue that the social network analysis community must confront. In this paper, we introduce a simplistic and effective machine learning based classifier to detect spammers. A multi-layer perceptron (MLP) with backpropagation training constitutes the core of the classifier. Using a public dataset of real-world social networks, the experiments demonstrated the possibility of reaching high accuracy levels with no more than two hidden layers.

Citation

Hemza Loucif , , (2025-06-06), A simplistic model for spammers detection in social recommender systems, Int. J. Business Information Systems, Vol:49, Issue:2, pages:199-217, Inderscience Enterprises Ltd

Detecting malicious user accounts on Twitter has become an active area of research in social network analysis. This kind of ill-intentioned users send undesired tweets to other users to promote products, services, rumors, fake news, or any abusive content. Hence, the detection of those spammers and their originators will prevent deterioration in the quality of communication services and legitimate users from being affected. Traditional machine learning techniques have been proposed to tackle the problem of spammers detection. However, many researchers have pointed out that the majority of machine learning based models that rely on supervised classification didn’t perform well in noisy and short message platforms like Twitter. Recently, deep learning-based alternatives have shown remarkable performance in this area because of their competitive training speed and low implementation cost. In this paper, we propose a new hybrid architecture that combines Principal Component Analysis (PCA) with Convolutional Neural Network (CNN) to give birth to a more reliable and robust model for spammers detection in Twitter. Unlike other hybridizations, the convolutional layer in the CNN module is not fed traditionally by raw feature vectors, rather, we use very low dimensional vectors containing high-order features provided by PCA module. A series of nicely conducted experiments over benchmark datasets have shown that the hybridization proved to be effective for the detection of spammers. The results show that PCA-CNN model can achieve better classification performance with 94.91% precision, 96.76% recall, and 95.83% F-score when compared to baseline benchmarks like CNN, ANN and SVM.

Citation

Hemza Loucif , , (2024-02-27), A Hybrid Deep Learning Approach for Spam Detection in Twitter, Ingénierie des Systèmes d’Information, Vol:29, Issue:1, pages:117-123, IIETA

This chapter presents a new version of one of the models that we have proposed to measure the influence of web users on social networks like Facebook. The new version enhanced the previous one through the incorporation of two substantial modules, namely, the global impression that the postings of the potential influencer get from his followers and the entropy which manifests the quantity of information carried by those postings. The comparison of our model with the precedent version and the PageRank benchmark has shown the effectiveness of our updates and the importance of incorporating the entropy and the global impression factors in its formulation.

Citation

Hemza Loucif , Samir Akhrouf , ,(2022-12-03), Toward a New Recursive Model to Measure Influence in Subscription Social Networks: A Case Study Using Twitter,International Conference on Managing Business Through Web Analytics,Algeria

This paper aims to describe a new simplistic model dedicated to gauge the online influence of Twitter users based on a mixture of structural and interactional features. The model is an additive mathematical formulation which involves two main parts. The first part serves to measure the influence of the Twitter user on just his neighbourhood covering his followers. However, the second part evaluates the potential influence of the Twitter user beyond the circle of his followers. Particularly, it measures the likelihood that the tweets of the Twitter user will spread further within the social graph through the retweeting process. The model is tested on a data set involving four kinds of real-world egocentric networks. The empirical results reveal that an active ordinary user is more prominent than a non-active celebrity one. A simple comparison is conducted between the proposed model and two existing simplistic approaches. The results show that our model generates the most realistic influence scores due to its dealing with both explicit (structural and interactional) and implicit features.

Citation

Hemza Loucif , , (2016-04-04), A simplistic model for identifying prominent web users in directed multiplex social networks: a case study using Twitter networks, New Review of Hypermedia and Multimedia, Vol:22, Issue:4, pages:287-302, taylor and francis