Editor’s note: This article originally appeared in the 2021 report on AI and Machine Learning in emerging media markets which was jointly prepared by International Media Support, The Fix, and El Clip. You can read the full report here.

Kyrgyzstan-based Kloop is known for its journalism investigations into corruption of Kyrgyz political elites and cases of electoral fraud on behalf of Kyrgyz state officials. Despite the modest team size of 30 full time equivalent staff, Kloop uses several AI/ML technologies – including graph databases, embedding and image recognition. The main reason for their data savvy strategy is based on the previous experience using it by its team members.

Logo of Kloop. Via kloop.kg 

These technologies enable the newsroom to find content insights. “Usually you need a cue from someone to start an investigation, but here you literally find stories inside the data,” says Kloop co-founder Rinat Tuhvatshin. Using this approach, the media discovered that the second-largest company was owned by a member of the parliamentary fuel committee. There are four directions in which Kloop applies its AI/ML efforts.

More from The Fix:  The next disruption: AI and Machine Learning in emerging market newsrooms

Investigations – the team use internally built algorithms to analyse and cross- reference large amounts of open-source data (legal entities, public procurement, tax declarations) to locate abnormal patterns with corruption risks. TV monitoring – Kloop records important TV channels’ broadcasts to process the recordings with image recognition tools and calculate coverage rates for pro-government and opposition politicians.

Election monitoring in real-time – Kloop developed a platform for its election monitors to enter violation reports. They were automatically wired to partner journalists and lawyers for examination. The platform processed the inputs and turned them into official appeals that a monitor could file in the polling station. Reader feedback – The Kyrgyz Political Compass tests readers’ views, helping them determine which political party is the closest to their views.

Kloop’s approach has included effective use of internships, using external resources for their own data collection needs. Internships can be a time-consuming task. “If you had 200 students to give each one a practical task, you would need 40 editors doing almost a full-time job”, explains Tuhvatshin. 

Instead, Kloop assigned the interns to collect data for a set of interviews with local district governors and wire it back to the data management system. “In this way, we were able to get interesting journalism material and provide tasks for 200 students and it took only one programmer to run it”.

More from The Fix: Creating projects of the future: Kyrgyz media hackathon case study

The 2021 report entitled ‘The next wave of Disruption: Emerging market media use of artificial intelligence and machine learning’ focuses on publishers applying AI, ML and Data Processing as part of their business operations in 20 countries of Latin America, and Central and Eastern Europe. The data was gathered via deep-dive case studies of 44 media outlets and over 33 hours of interviews with experts.

Photo by Mike Dudin on Unsplash