Entity linking in Arabic text presents unique challenges due to complex morphology, absence of capitalization, and dialectal variations. Traditional approaches rely on static entity representations that fail to capture contextual dynamics. This paper introduces SAGE (Semantic Adaptation of Graph Embeddings), a novel framework for Arabic entity linking that dynamically updates entity representations through a Graph Attention Network (GAT) architecture. SAGE integrates AraBERTv2 contextual embeddings with a hierarchical GAT design that processes information at both entity and type levels. The framework’s key innovation lies in its dynamic vector update mechanism that propagates semantic changes through the knowledge graph. Evaluation on Arabic news articles demonstrates that SAGE outperforms static representation methods by $28 \%$ absolute improvement in linking accuracy, with particularly significant improvements for person entities and ambiguous mentions.

Nowadays, Wireless Multimedia Sensor Networks (WMSNs) plays a vital role in modern surveillance systems and enables innovative solutions across diverse sectors. In such networks, detecting and minimizing data redundancies has become crucial for both prolonging network lifespan and improving data quality. In this paper, we propose a two-layer reduction scheme, called as SHOLO, that efficiently detects and removes similarities in SHOrt and LOng term video data collected in WMSNs, thus conserve sensor energies and extending network lifetime. In the first layer, SHOLO detects and removes similarities among successive frames captured by video node during the same period time through two novel distance-based methods: Difference Hash (DH) and Pixel Intensity Threshold (PIT). In the second layer, SHOLO searches long-term similarities among frames collected in the same period to detect scene variation and zone dynamicity. Then, we introduced two similarity reduction methods in such layer: Intensity-based Difference (ID) and Block-based Difference (BD). We conducted extensive simulations using real-world video data set to validate the efficiency of the proposed framework. The results demonstrated that SHOLO can reduce up to 93.8% of collected video data, leading to tremendous energy savings and network lifetime compared to existing approaches.

Sentiment analysis involves using computational methods to identify and classify opinions expressed in text, with the goal of determining whether the writer's stance towards a particular topic, product, or idea is positive, negative, or neutral. However, sentiment analysis in Arabic presents unique challenges due to the complexity of Arabic morphology and the variety of dialects, which make language classification even more difficult. To address these challenges, we conducted to investigation and overview the techniques used in the last five years for embedding and classification of Arabic sentiment analysis (ASA). We collected data from 100 publications, resulting in a representative dataset of 2,300 detailed records that included attributes related to the dataset, feature extraction, approach, parameters, and performance measures. Our study aimed to identify the most powerful approaches and best model settings by analyzing the collected data to identify the significant parameters influencing performance. The results showed that Deep Learning and Machine Learning were the most commonly used techniques, followed by lexicon and transformer-based techniques. However, Deep Learning models were found to be more accurate for sentiment classification than other Machine Learning models. Furthermore, multi-level embedding was found to be a significant step in improving model accuracy.

In order to improve link prediction process in a knowledge graph, we tackle the problem of learning representations of entities and relations. The strength of the existing models in this domain mainly relies on its ability of modeling and inferring the patterns of the relations. In this paper, we propose a new knowledge graph embedding approach called TransModE that has the ability to represent all simple and complex relation patterns. Inspired by TransE, TransModE represents the relations between the entities of a knowledge graph as a transition in the modulus space. Experimenting our model on multiple benchmark knowledge graphs shows the simplicity and the scalability of TransModE. It also shows that TransModE outperforms existing state-of-the-art models by being able to infer and model all types of relations.

Knowledge graph is a knowledge representation technique that helps representing entities and relations in a machine understandable way.This promising trend suffers from the problem of incompleteness that was best solved by link prediction.Indeed, link prediction is the most successful method for understanding the structure of the large knowledge graphs.Knowledge graph embedding KGE is one of the best link prediction methods.Its effectiveness is mainly affected by the accuracy of learning representations of entities and relations.In this paper, we propose a new knowledge graph embedding model called KEMA ( Knowledge-graph Embedding using Modular Arithmetic).KEMA has the ability to represent simple and complex relations in an efficient way.Consequently, this allows our model to outperform the majority of the existing models.Mainly, KEMA depends on representing the relations in a knowledge graph by modular arithmetic operations applied between entities.Experimental results on multiple benchmark knowledge graphs verify the accurate representation, low complexity and scalability of KEMA.

Nowadays, representing entities and relations in a machine understandable way through Knowledge graph embedding (KGE) has been proven as an effective approach for predicting missing links in knowledge graphs (KGs). Mainly, the success of such approach depends on the model ability to infer the patterns of the relations. Indeed, most of the existing KG models highly focus on modeling simple relation patterns such as symmetry, anti-symmetry, inversion, and composition. However, there are few models in the literature that take into consideration the modeling of complex relation patterns like 1-[Formula: see text], [Formula: see text]-1 and [Formula: see text]-[Formula: see text], which are common in real-world applications. To overcome this challenge, this paper presents a new KGE model called KEMA[Formula: see text], i.e. KGE using Modular Arithmetic, that relies on the combination of projection and modular arithmetic. The main idea behind KEMA[Formula: see text] is to project the entities of a relation to represent the relations of a KG, before applying a modular arithmetic over it. Thus, KEMA[Formula: see text] will be able to infer all simple and complex relation patterns for any KGE applications. Through extensive experiments on several datasets, we demonstrated the relevance of KEMA[Formula: see text] in terms of effectively representing all model relations in a KG. Simulations on several tested datasets show that KEMA[Formula: see text] obtains the good scores for Mean Rank (MR) and Hits@1 tests. Moreover, KEMA[Formula: see text] obtains the good Hits@1 score compared to the existing models.

Search engines are among the most popular web services on the World Wide Web. They facilitate the process of finding information using a query-result mechanism. However, results returned by search engines contain a lot of duplications. For instance, when a user searches for a query q on Google, he will get n number of links divided into m number of pages. Although returned links have different content, exactly similar content is still existing on different links returned by search engines. This problem is referred to as duplication. Solving this problem, will increase the quality of search results as well as reducing the time search per query. In this paper, we introduce a new method called REDUCE (REsult DUplication detection in searCh Engines), to address this problem. It implements a semi-supervised approach. It approximately measures the similarity between the web pages and we suggest a new method to group the search results based on their similarity. To evaluate our method, we collect data from Google and other search engine platforms. We show that our method can solve this problem on different search engine platforms with different languages. We empirically evaluated our results on different classification algorithms and reached an accuracy of 96.7%.

Many authors can share the same name and this constitutes a serious problem that affects the relevancy of retrieval results and constitutes our motivation of finding such approach to cover this issue at the author names entity level. Solving such a problem may return with positive gain at the level of document retrieval, web search and the quality of data. This entity resolution task can be tackled as an unsupervised problem, where there are set of features that can be employed for the resolution job, or as supervised problem to compute the similarities among two citations and then classify if they are the same or not. Recent approaches usually utilize features such as: co-author, venue, topic similarity, affiliations and title of publications to deal with author ambiguity. In this paper, three attributes are used to treat this problem sequentially. The co-authorship firstly which is a well-known attribute, and then the topic and affiliation extracted from biographies, which can be found inside the publication, and this is our novelty frame in this paper.

The accelerating progress in science with the active role of the communication media -mainly the web -make person in front of a difficult task, in finding appropriate information during a brief time.In a narrower context, many researches were created in the expertise retrieval domain, as an interesting and complicated task for the scientific community, in face of this huge amount of data scattered across the web.Benefiting from the semantic web technologies and the efforts of data structuring, in this paper we propose a novel approach of correlation based profile building, by exploiting heterogynous web sources.The aim is to generate comprehensive and validated profiles about researchers and experts in the computer science domain.