Article: An entity matching-based image topic verification framework for online fact-checking Journal: International Journal of Multimedia Intelligence and Security (IJMIS) 2022 Vol.4 No.1 pp.65 - 85 Abstract: The last decade has witnessed an unprecedented growth in online multimedia data. However, the manipulated and fake images have created fertile grounds for sowing online fake news. Consequently, online fact-checking has drawn more attention from academia and industry to detect and mitigate online fake news. Nevertheless, most of the online fact-checking task focus on textual content. Although multimedia information like images can provide promising potentials for identifying misinformation, it has not been adequately studied. Besides, traditional information retrieval techniques, e.g., image caption generation, typically lack high-quality training data or their computation costs are very high. Aiming to address the above issues, we proposed an image topic verification framework based on named entity matching. Particularly, the proposed framework can effectively check if a targeted image is related to a specific topic or not. In addition, it can also retrieve helpful contextual background and knowledge about the targeted image. We conduct extensive experiments and analyses. The results validate the effectiveness and practicality of our framework. Inderscience Publishers - linking academia, business and industry through research

Title: An entity matching-based image topic verification framework for online fact-checking

Authors: Xichen Zhang; Sajjad Dadkhah; Samaneh Mahdavifar; Rongxing Lu; Ali A. Ghorbani

Addresses: Faculty of Computer Science, Canadian Institute for Cybersecurity, University of New Brunswick, Fredericton, NB, Canada ' Faculty of Computer Science, Canadian Institute for Cybersecurity, University of New Brunswick, Fredericton, NB, Canada ' Faculty of Computer Science, Canadian Institute for Cybersecurity, University of New Brunswick, Fredericton, NB, Canada ' Faculty of Computer Science, Canadian Institute for Cybersecurity, University of New Brunswick, Fredericton, NB, Canada ' Faculty of Computer Science, Canadian Institute for Cybersecurity, University of New Brunswick, Fredericton, NB, Canada

Abstract: The last decade has witnessed an unprecedented growth in online multimedia data. However, the manipulated and fake images have created fertile grounds for sowing online fake news. Consequently, online fact-checking has drawn more attention from academia and industry to detect and mitigate online fake news. Nevertheless, most of the online fact-checking task focus on textual content. Although multimedia information like images can provide promising potentials for identifying misinformation, it has not been adequately studied. Besides, traditional information retrieval techniques, e.g., image caption generation, typically lack high-quality training data or their computation costs are very high. Aiming to address the above issues, we proposed an image topic verification framework based on named entity matching. Particularly, the proposed framework can effectively check if a targeted image is related to a specific topic or not. In addition, it can also retrieve helpful contextual background and knowledge about the targeted image. We conduct extensive experiments and analyses. The results validate the effectiveness and practicality of our framework.

Keywords: image topic verification; fact-checking; named entity matching.

DOI: 10.1504/IJMIS.2022.121286

International Journal of Multimedia Intelligence and Security, 2022 Vol.4 No.1, pp.65 - 85

Received: 20 Sep 2021
Accepted: 29 Nov 2021
Published online: 03 Mar 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: An entity matching-based image topic verification framework for online fact-checking

Keep up-to-date