‘AI Troubleshooter Startup’ Emerges Predicting Ratings and Recommending Actors

‘Dae-Star Problem Solver Platform’ KBS AI-Based Drama Viewership Prediction Contest Results
Didive, CoreDotToday, Tendy Selected... Up to 2.5 Billion KRW Government Support

Park Young-sun, Minister of SMEs and Startups, is delivering a greeting at the Ministry of SMEs and Startups Daestar Solver Platform Performance Sharing Conference held on the afternoon of the 16th at KT Square in Jongno-gu, Seoul. Photo by Yonhap News

[Asia Economy Reporter Kim Heeyoon] The Ministry of SMEs and Startups announced on the 21st that it held the finals for the Korea Broadcasting System (KBS) 'AI-based Drama Viewership Prediction' project on the Big-Star Solver Platform and selected three winning teams.

This project was planned to reduce production costs while improving broadcasters' competitiveness with good viewership ratings and to strengthen support for startups in the media sector utilizing AI.

The winning teams were Didive, CoreDotToday, and Tendy. Five startups participated in the finals. They presented the experimental results conducted in advance on KBS's daily drama 'Nuga Mworaedo' and Munhwa Broadcasting Corporation (MBC) mini-series 'Kairos,' the validity of the algorithms used to derive the results, and the practical applicability for future drama production decisions.

The evaluation was reviewed for usability by KBS, and Microsoft, the technical support company for the contest, along with AI experts, verified the validity and scalability of the algorithms.

Didive, which produced the closest results in the pre-experiment for drama viewership prediction, accurately predicted using an analysis of 11 variables utilizing Microsoft's Azure machine learning, receiving high praise from the judges.

CoreDotToday submitted an algorithm that derives genres from drama synopses and analyzes characters by role to predict drama viewership for candidate actors. It was evaluated as an algorithm that improves the accuracy of drama performance prediction through AI learning and assists in actor selection at the pre-planning stage.

Tendy analyzed success factors of dramas focusing on the combination of actors, topicality, production staff, and public interest online. They introduced an algorithm that extracts character factors of dramas and learns past viewership ratings of previous works and similarity between actors to recommend actors and actor combinations expected to increase viewership, attracting great interest in terms of ease of use.

Meanwhile, the final selected startups will be awarded at a performance sharing event on the 1st of next month. The selected companies will receive up to 2.5 billion KRW in government support, including commercialization funds, technology development funds, and technology special guarantees, along with opportunities to jointly promote projects with large corporations and expand globally using overseas networks.