Said Kerrache

Abdullah Alhadlaq, Said Kerrache, Hatim Aboalsamh

Online stores and service providers rely heavily on recommendation softwares to guide users through the vast amount of available products. Consequently, the field of recommender systems has attracted increased attention from the industry and academia alike, but despite this joint effort, the field still faces several challenges. For instance, most existing work models the recommendation problem as a matrix completion problem to predict the user preference for an item. This abstraction prevents the system from utilizing the rich information from the ordered sequence of user actions logged in online sessions. To address this limitation, researchers have recently developed a promising new breed of algorithms called sequence-aware recommender systems to predict the user's next action by utilizing the time series composed of the sequence of actions in an ongoing user session. This paper proposes a novel sequence-aware recommendation approach based on a complex network generated by the hidden metric space model, which combines node similarity and popularity to generate links. We build a network model from data and then use it to predict the user's subsequent actions. The network model provides an additional source of information that improves the accuracy of the recommendations. The proposed method is implemented and tested experimentally on a large dataset. The results prove that the proposed approach performs better than state-of-the-art recommendation methods.

Abdullah Alhadlaq, Said Kerrache, Hatim Aboalsamh

Recommender systems have become an essential tool for providers and users of online services and goods, especially with the increased use of the Internet to access information and purchase products and services. This work proposes a novel recommendation method based on complex networks generated by a similarity-popularity model to predict ones. We first construct a model of a network having users and items as nodes from observed ratings and then use it to predict unseen ratings. The prospect of producing accurate rating predictions using a similarity-popularity model with hidden metric spaces and dot-product similarity is explored. The proposed approach is implemented and experimentally compared against baseline and state-of-the-art recommendation methods on 21 datasets from various domains. The experimental results demonstrate that the proposed method produces accurate predictions and outperforms existing methods. We also show that the proposed approach produces superior results in low dimensions, proving its effectiveness for data visualization and exploration.

Said Kerrache, Yasushi Nakauchi

Optimal mass transport, also known as the earth mover's problem, is an optimization problem with important applications in various disciplines, including economics, probability theory, fluid dynamics, cosmology and geophysics to cite a few. Optimal transport has also found successful applications in image registration, content-based image retrieval, and more generally in pattern recognition and machine learning as a way to measure dissimilarity among data. This paper introduces the problem of constrained optimal transport. The time-dependent formulation, more precisely, the fluid dynamics approach is used as a starting point from which the constrained problem is defined by imposing a soft constraint on the density and momentum fields or restricting them to a subset of curves that satisfy some prescribed conditions. A family of algorithms is introduced to solve a class of constrained saddle point problems, which has convexly constrained optimal transport on closed convex subsets of the Euclidean space as a special case. Convergence proofs and numerical results are presented.

Ghadah Alabduljabbar, Hafida Benhidour, Said Kerrache

Image captioning is the process of automatically generating a description of an image in natural language. Image captioning is one of the significant challenges in image understanding since it requires not only recognizing salient objects in the image but also their attributes and the way they interact. The system must then generate a syntactically and semantically correct caption that describes the image content in natural language. With the significant progress in deep learning models and their ability to effectively encode large sets of images and generate correct sentences, several neural-based captioning approaches have been proposed recently, each trying to achieve better accuracy and caption quality. This paper introduces an encoder-decoder-based image captioning system in which the encoder extracts spatial features from the image using ResNet-101. This stage is followed by a refining model, which uses an attention-on-attention mechanism to extract the visual features of the target image objects, then determine their interactions. The decoder consists of an attention-based recurrent module and a reflective attention module, which collaboratively apply attention to the visual and textual features to enhance the decoder's ability to model long-term sequential dependencies. Extensive experiments performed on Flickr30K, show the effectiveness of the proposed approach and the high quality of the generated captions.

Hafida Benhidour, Lama Almeshkhas, Said Kerrache

Complex networks are graphs representing real-life systems that exhibit unique characteristics not found in purely regular or completely random graphs. The study of such systems is vital but challenging due to the complexity of the underlying processes. This task has nevertheless been made easier in recent decades thanks to the availability of large amounts of networked data. Link prediction in complex networks aims to estimate the likelihood that a link between two nodes is missing from the network. Links can be missing due to imperfections in data collection or simply because they are yet to appear. Discovering new relationships between entities in networked data has attracted researchers' attention in various domains such as sociology, computer science, physics, and biology. Most existing research focuses on link prediction in undirected complex networks. However, not all real-life systems can be faithfully represented as undirected networks. This simplifying assumption is often made when using link prediction algorithms but inevitably leads to loss of information about relations among nodes and degradation in prediction performance. This paper introduces a link prediction method designed explicitly for directed networks. It is based on the similarity-popularity paradigm, which has recently proven successful in undirected networks. The presented algorithms handle the asymmetry in node relationships by modeling it as asymmetry in similarity and popularity. Given the observed network topology, the algorithms approximate the hidden similarities as shortest path distances using edge weights that capture and factor out the links' asymmetry and nodes' popularity. The proposed approach is evaluated on real-life networks, and the experimental results demonstrate its effectiveness in predicting missing links across a broad spectrum of networked data types and sizes.

Said Kerrache, Hafida Benhidour

Graph embedding methods aim at finding useful graph representations by mapping nodes to a low-dimensional vector space. It is a task with important downstream applications, such as link prediction, graph reconstruction, data visualization, node classification, and language modeling. In recent years, the field of graph embedding has witnessed a shift from linear algebraic approaches towards local, gradient-based optimization methods combined with random walks and deep neural networks to tackle the problem of embedding large graphs. However, despite this improvement in the optimization tools, graph embedding methods are still generically designed in a way that is oblivious to the particularities of real-life networks. Indeed, there has been significant progress in understanding and modeling complex real-life networks in recent years. However, the obtained results have had a minor influence on the development of graph embedding algorithms. This paper aims to remedy this by designing a graph embedding method that takes advantage of recent valuable insights from the field of network science. More precisely, we present a novel graph embedding approach based on the popularity-similarity and local attraction paradigms. We evaluate the performance of the proposed approach on the link prediction task on a large number of real-life networks. We show, using extensive experimental analysis, that the proposed method outperforms state-of-the-art graph embedding algorithms. We also demonstrate its robustness to data scarcity and the choice of embedding dimensionality.

Reem Alothman, Hafida Benhidour, Said Kerrache

The widespread use of text-based communication on social media-through chats, comments, and microblogs-has improved user interaction but has also led to an increase in offensive content, including hate speech, racism, and other forms of abuse. Due to the enormous volume of user-generated content, manual moderation is impractical, which creates a need for automated systems that can detect offensive language. Deep learning models, particularly those using transfer learning, have demonstrated significant success in understanding natural language through large-scale pretraining. In this study, we propose an automatic offensive language detection model based on XLNet, a generalized autoregressive pretraining method, and compare its performance with BERT (Bidirectional Encoder Representations from Transformers), which is a widely used baseline in natural language processing (NLP). Both models are evaluated using the Offensive Language Identification Dataset (OLID), a benchmark Twitter dataset that includes hierarchical annotations. Our experimental results show that XLNet outperforms BERT in detecting offensive content and in categorizing the types of offenses, while BERT performs slightly better in identifying the targets of the offenses. Additionally, we find that oversampling and undersampling strategies are effective in addressing class imbalance and improving classification performance. These findings highlight the potential of transfer learning and XLNet-based architectures to create robust systems for detecting offensive language on social media platforms.

Hanan S. Murayshid, Hafida Benhidour, Said Kerrache

This paper proposes a sequence-to-sequence learning approach for Arabic pronoun resolution, which explores the effectiveness of using advanced natural language processing (NLP) techniques, specifically Bi-LSTM and the BERT pre-trained Language Model, in solving the pronoun resolution problem in Arabic. The proposed approach is evaluated on the AnATAr dataset, and its performance is compared to several baseline models, including traditional machine learning models and handcrafted feature-based models. Our results demonstrate that the proposed model outperforms the baseline models, which include KNN, logistic regression, and SVM, across all metrics. In addition, we explore the effectiveness of various modifications to the model, including concatenating the anaphor text beside the paragraph text as input, adding a mask to focus on candidate scores, and filtering candidates based on gender and number agreement with the anaphor. Our results show that these modifications significantly improve the model's performance, achieving up to 81% on MRR and 71% for F1 score while also demonstrating higher precision, recall, and accuracy. These findings suggest that the proposed model is an effective approach to Arabic pronoun resolution and highlights the potential benefits of leveraging advanced NLP neural models.