Preface of the special issue on advances in data-driven engineering

Disruptive changes in the industrial environment have occurred in recent years owing to rapid advancements in electronics, information, and communication technology. Because of the ever-increasing demands for product quality and economic benefit, intelligent components and devices are implemented, and networked and real-time supervision and control systems are also running in parallel. As a consequence, the level of automation in modern industrial systems is steadily rocketing. In addition, the increased availability of different data types paves the way to stunning scenarios for applying data-driven modeling techniques. The latter are revolutionizing complex systems' modeling, prediction, and control. Fresh advances in scientific computing witness how data-driven methods can be applied to diverse, complex systems. Applications of Artificial intelligence-based systems play a pivotal role at the crossroads of almost all fields of computer science. Recent advances in big data generation and management have allowed decision-makers to utilize these overwhelming volumes of data for various purposes and analyses.

This special issue consists of selected papers from an open call as well as thoroughly revised papers from the 2021 International Conference on Model and Data Engineering (MEDI'2021) held remotely in Tallinn (Estonia) (Attiogbe and Ben Yahia 2021). This special issue unveils new trends in developing data-driven application systems that seek to adapt computational algorithms and techniques in many application domains, including software systems, cyber security, human activity recognition, and behavioural modeling. Original research and review work with models and building data-driven applications using computational algorithms were particularly sought after.

This special issue, aiming to provide state-of-the-art information to academics, researchers, and industry practitioners on Advances in Data-driven Engineering, attracted a total of eleven (11) submissions, five (5) of which had their initial versions among the sixteen (16) full papers presented during the MEDI'2021 conference. The remaining articles are contributions submitted in response to the general call for the special issue. Among the eleven submitted papers, the following six (6) papers were accepted after a thorough two-level reviewing process.

The first paper in this special issue is authored by Garcia-Garcia et al. (2023). The authors introduced the design and the implementation of efficient distributed algorithms for distance join queries in Spark-based spatial analytics systems. They look into how to make and use efficient distance-based join queries and distributed algorithms in Apache Sedona. The authors improved the new in-memory cluster computing system for processing large-scale spatial data using the best spatial partitioning and other optimization techniques.

In the second paper, Ellouze, Mechtib, and Belguith (2023) proposed a supervised learning method leveraging multimodal information for paranoid detection in French tweets during the COVID-19 outbreak. Statistical techniques combined and filtered the features extracted from the processing steps to gauge the distance between the training corpus (the labeled data) and the test corpus (the unlabeled data). The authors report very encouraging results, with an accuracy of 78% for the detection of paranoid people and 70% for the detection of the behaviour of these people toward COVID-19.

Identifying “key” nodes in social networks that are substantial for information spread and control is the main purpose of the third paper authored by Jain and Sinha (2023). The latter introduced TriBeC, a novel centrality metric, to identify the significant nodes in online social networks by utilizing the impact of weighted betweenness extended with network quartiles. Experimental validation outcomes on Twitter, Facebook, BlogCatalog, scale-free, and random networks show the outperforming results of the topmost 1% TriBeC central nodes over their existing counterparts in terms of the percentage of the network being infested with information over time.

The fourth paper, authored by Berkani et al. (2023), proposed a novel recommendation model called S-SNHF (Sentiment-based Social Neural Hybrid Filtering). Using a deep neural architecture, this model combines collaborative and content-based filtering with social information. The results of the empirical study performed show that the introduced approach outperforms its competitors and achieves significantly better recommendation accuracy.

Vovk, Piho, and Ross (2023), in the fifth paper, pay heed to the utmost issue of privacy violations in the ever-evolving Healthcare field. The authors scrutinized the scientific literature on various methods and tools for anonymization published between 2017 and 2021. The ultimate goal is an insightful and comprehensive overview of the methods used in health data anonymization and the available tools that use those methods.

Charbel et al. (2023) authored the last paper of this special issue. They introduced Feed2Search, a novel framework for hybrid molecule-based semantic search, which facilitates information retrieval over a heterogeneous document corpus. In addition to a semantic representation of the corpus, the authors propose a query processing pipeline based on a novel data structure for query answers extracted from this graph, which embeds core information with structural-based and domain-specific contexts. The experiments pinpoint promising results in real-world construction projects from the Architecture, Engineering, and Construction industries and show promising results and sketch issues for further investigation in other domains.

We would like to take this opportunity to acknowledge the tireless cooperation and support of the reviewers, even if deadlines were tight. All submissions were reviewed by at least three reviewers from a pre-eminent panel of reviewers that worked diligently to guarantee a thorough review of each paper.

In closing, we would like to express our gratitude to the authors who contributed to this ijgs special issue that will greatly interest everyone working in concept lattices and their applications. Last but not least, we are grateful to Radim Belohlavek, the ijgs Editor-in-Chief, who provided continuous support and advice while preparing this special issue.