Effective detection techniques of false or unverified information on the web is a contemporary theoretical and applied research topic, because of the growth of the malicious use of fake news and trolling. Early studies, specially from a computational perspective, are relatively recent and the boundaries of the study matter are often not clearly defined. Anyway, literature provides some main insights related to the type of false information and the detection techniques: besides well-known “fake news”, “rumors” and “blunders” from the web, literature defines also “clickbait”, “social spam” and “fake reviews”. Category mostly depends on sources and type of data used for analysis.
The most promising approaches for false information detection consider the task as a classification problem and focus on using content and context features for classification, but very few rely on systematic sources. The main classification approaches use machine learning and deep learning techniques; the last ones have obtained state-of-the-art results and most recent approaches are focused on exploiting such frameworks to some extent. Finally, a major challenge is the lack of widely accepted benchmark datasets and accepted trusted sources lists. This is fundamental to compare and evaluate the effectiveness of each approach. Currently available resources may not be sufficient for gaining novel insight on relevant properties of false information and building models able to operate in a real world scenario.
This Invited Session aims to call novel contributions and recent advances from the research communities to provide a support in helping to pave the way for a new culture of information, by the means of methodological research for methods and technological instruments for combating cultural misinformation and mitigating the wide and viral spreading of information blunders that causes false knowledge acquisition by social communities and serious and permanent damages in the social behavior of people.
Both methodological and applicative works are welcome to this Session, whose aim is, among the others, the attempt to build an interdisciplinary research group, made of linguistic, communication, statistics, security and computer science, from the international research community, in order to reach the identification of a methodological framework for supporting automatic but effective classification of trusted information sources.
We encourage novel research contribution highlighting outcomes in validation and verification processes and quantitative evaluation metrics for information source ranking.
Artificial Intelligence techniques and methods, joint with advanced statistical methods for multidimensional data analysis and cybersecurity warnings and case studies represent fundamental building blocks.
Fiammetta Marulli, Università degli Studi della Campania "L. Vanvitelli", fiammetta.marulli@unicampania.it
Lelio Campanile, Università degli Studi della Campania "L. Vanvitelli", lelio.campanile@unicampania.it
Laura Verde, Università degli Studi della Campania "L. Vanvitelli", laura.verde@ unicamapania.it
The Full Papers conference proceedings will be published by Springer as book chapters in a volume of the KES Smart Innovation Systems and Technologies series, submitted for indexing in Scopus and Thomson-Reuters Conference Proceedings Citation Index (CPCI) and the Web of Science.
To submit a paper please follow the procedure described at the KES-IDT-22 website http://idt-22.kesinternational.org/submission.php.