Social Media and Web Multimedia Mining

The pervasive use of Online Social Networks (OSN) for networking, communication and search in tandem with the ubiquitous availability of smartphones, which enables real-time multimedia capturing and sharing, have led to massive amounts of user-generated content and activities being amassed online, and made publicly available for analysis and mining. Leveraging large amounts of user-generated content creates new exciting possibilities for a number of applications and services such as search and forecasting. Yet, processing such content involves facing a number of hard challenges stemming from its scale and nature, while there are also a number of important caveats when dealing with user-generated content, such as the inference and disclosure of personal information. As a result, conventional multimedia analysis methods are not sufficient to extract value from social media and web content. To this end, MKLab develops methods to tackle a number of pertinent problems:

  • efficient processing and indexing algorithms that can handle massive amounts of content;
  • distributed and streaming data management frameworks to deal with the real-time nature of social media content;
  • novel retrieval models that take into account both relevance and diversity;
  • intuitive and engaging visualizations and dashboards that capture and summarize the key topics and interactions of online communities;
  • machine learning frameworks that help uncover malicious and false user-generated information;
  • graph-based algorithms that analyze the social context and interactions among OSN users;
  • privacy-oriented data management frameworks that help OSN users gain awareness and control with respect to personal information disclosure and privacy risks;
  • social media-powered applications that leverage social media content for improving performance in a given task, e.g. predicting the election result.

During the recent years, our lab has been particularly active in the field of media-based disinformation, developing methods and tools for the (semi-) automatic detection and characterization of misleading and tampered content in online settings (web pages, social media), including image forensics, fake tweet post classification, video verification, etc. Due to the breadth of activities on this area, we have recently launched a dedicated website with more up-to-date and extensive information on the relevant projects and outcomes.

Key publications

  1. E. Spyromitros-Xioufis, S. Papadopoulos, A. Ginsca, A. Popescu, Y. Kompatsiaris, I. Vlahavas, “Improving Diversity in Image Search via Supervised Relevance Scoring”, Proc. International Conference on Multimedia Retrieval (ICMR), Shanghai, China, June 23-26, 2015.

  2. A. Tsakalidis, S. Papadopoulos, A. Cristea, Y. Kompatsiaris. “Predicting Elections for Multiple Countries Using Twitter and Polls”, IEEE Intelligent Systems.

  3. S. Papadopoulos, Y. Kompatsiaris, “Social Multimedia Crawling for Mining and Search”, IEEE Computer, Volume 47, Issue 5, May 2014, pp. 84-87

  4. L. Aiello, G. Petkos, C. Martin, D. Corney, S. Papadopoulos, R. Skraba, A. Goker, I. Kompatsiaris, A. Jaimes, “Sensing trending topics in Twitter”, Transactions on Multimedia, online preprint, IEEE, 2013, doi:10.1109/TMM.2013.2265080

  5. S. Papadopoulos, Y. Kompatsiaris, A. Vakali, P. Spyridonos. “Community Detection in Social Media”. In Data Mining and Knowledge Discovery, Vol. 24, Issue 3, pp. 515-554, May 2012

  6. Schinas, M., Papadopoulos, S., Petkos, G., Kompatsiaris, Y., & Mitkas, P. A. (2015, Oct). “Multimodal Graph-based Event Detection and Summarization in Social Media Streams”. In Proceedings of the 23rd Annual ACM Conference on Multimedia Conference (pp. 189-192). ACM.

  7. Andreadou, K., Papadopoulos, S., Apostolidis, L., Krithara, A., & Kompatsiaris, Y. (2015). “Media REVEALr: A Social Multimedia Monitoring and Intelligence System for Web Multimedia Verification”. In Intelligence and Security Informatics (pp. 1-20). Springer International Publishing.