Multimedia Data Mining and Processing for News Source Attribution

Abstract

The desire for unbiased journalism that effectively counters disinformation is widely recognised. News consumers are not only interested in news, but they also want unbiased journalism that cuts through disinformation, and they want it from trusted news sources. Consequently, media researchers need to explore ways to facilitate news-source identification, irrespective of the platform used. However, the availability of multimedia data sources has seen a remarkable surge in recent years, encompassing demographic data, social media data, geodata, and pervasive digital trace data. Multimedia data mining is a procedure of discovering stimulating trends via media data using video, text, and audio that are not generally available by simple enquiries and related outputs. Researchers face the challenge of integrating these diverse sources to enhance news source attribution in multimedia data including platforms like Facebook, WhatsApp and Instagram. The paper presents a review of multimedia data approaches and their application to news source attribution research. Also, the examination of the benefits and limitations of these techniques and discussion on future directions were mentioned. Consideration was on machine learning and statistical approaches to multimedia data, which include deep learning, and probabilistic modelling. Similarly, a discussion on the importance of data privacy and ethics in news source attribution research was stated. The contribution of this study is highly relevant for news media research groups striving to improve their capability to attribute sources in multimedia data, thereby combatting disinformation and amplifying trusted media brands. Keywords: Data mining, Multimedia, Data process, News source attribution, Unbiased journalism