An interview with Mr. Georgi Kostadinov, Head of Artificial Intelligence (AI), Imagga Technologies Ltd, Bultaria: “TCounteR is not just a research project – it is a tool to be used in real life for understanding and combating radicalisation online“, published in CounteR newsletter #2.
Dear Mr. Georgiev, what is Imagga’s role in CounteR?
Imagga is contributing with the development of instruments and deep learning algorithms for analysis of images, used for automatic detection of radicalisation and extremist content. We are leveraging Imagga’s image recognition expertise and proprietary technology to build robust mechanisms for detecting problematic content. This includes detecting specific keywords, a scene in an image or action in a video that might be related or incline to radicalisation.
Would you share some updates on the work’s progress?
We have started the efforts to collecting training data for the development of AI algorithms. The input data is crucial for any deep learning model, and having access to sensitive and very specific data via some of the project partners is the only way to tackle the complexity of the task and ensure the image recognition technology will be able to detect and flag problematic content. We are also actively working on Work Package 3 (see Newletter’s front page for more info on work packages) and more specifically for Instagram Data Collector. Internally, we are making efforts to specify Imagga’s content moderation technology for radicalised content.
What results do you anticipate for Imagga from its participation in CounteR?
The most important outcome of our participation in the project is to further develop Imagga’s content moderation technology to better detect radicalised content. All participants in the project come with their specific knowledge and industry experience needed for building a complex and multi-disciplinary project such as CounteR. Another benefit is the direct access to LEAs, who are the end-users of the CounteR technology.
What themes would you be interested in discussing with industry stakeholders through CounteR’s Cluster Group?
Some of the themes that would be of great interest to Imagga are: how to better understand what radicalised content is; what are the triggers for radicalisation and how they can be detected; how to improve the effectiveness of early detection and prevent such content from becoming viral; and finally what efforts and methods other research groups and businesses are using to effectively detect problematic content related to radicalisation.
How can CounteR ensure that its research will meet the needs of the end-users?
One of the best ways is to include LEAs and end-users in the process of drafting and developing the actual product. Doing expert interviews, learning from the LEAs’ field expertise and implementing that knowledge will ensure that CounteR is not just a research project, but a tool to be used in real life for understanding and combating radicalisation online. The technical requirements should be in line with what LEAs have in mind and need to effectively fight radicalisation.
Georgi Kostadinov is Head of Artificial Intelligence (AI) at Imagga Technologies Ltd./Bulgaria. He has been recognised by Forbes Bulgaria in 2020 as one of the 30 people under the age of 30 for achievements in the areas of computer science and innovation. Mr. Kostadinov has architected and implemented the largest commercial plant recognition model with 100K+ categories, 150M sample images, and 97% top-5 accuracy.