Kibo, Establishing an 'AI-Based New Technology Evaluation Model'

Applying AI Algorithms to Over 710,000 Technology Evaluation Data, Improving Prediction Accuracy by More Than 9%

[Asia Economy Reporter Kim Cheol-hyun] The Korea Technology Finance Corporation (Chairman Jung Yoon-mo, hereinafter referred to as Kibo) announced on the 23rd that it has completed the development of a new artificial intelligence (AI)-based technology evaluation model through a self-assessment of the technology evaluation model previously applied and will sequentially apply it to technology evaluations starting in the second half of the year.

Based on proactive research to introduce 4th industrial revolution technologies into technology finance, Kibo developed 'KPAS,' a system in which AI learns the evaluation patterns of expert evaluators to assess the value of patents. This time, AI algorithms were applied to the technology evaluation system that grades the value of technology through the new technology evaluation model.

Kibo explained that it analyzed a vast amount of technology evaluation data accumulated over the past 15 years, totaling approximately 710,000 cases, to standardize the technology evaluation model and improve the efficiency of technology evaluations. An internal expert group began diagnosis and research, completing the development of a new technology evaluation model with AI algorithms applied. The new technology evaluation model improved prediction accuracy by more than 6% for future growth potential and more than 12% for corporate insolvency risk compared to the existing model. Overall, the prediction accuracy increased by more than 9%.

Performance Comparison of Technology Evaluation Models

Furthermore, the model’s stability was secured by organically combining expert opinions with AI technology. The difficulty in interpreting evaluation item grade decisions, which had been an obstacle to AI adoption in the financial sector, was resolved by introducing AI interpretation techniques. In addition, the new technology evaluation model standardized the previously segmented various models, increasing the processing speed of technology evaluation tasks and shortening the development period for new models, enabling a swift response to diverse policy demands in crisis situations such as the COVID-19 pandemic.

Kibo plans to publish the development results of the new technology evaluation model in domestic and international professional journals to promote academic sharing and objective verification simultaneously, and will also strive to secure intellectual property rights through patent applications for the model. Chairman Jung Yoon-mo of Kibo stated, "Since developing the first domestic technology evaluation model in 2005, Kibo has pioneered a new horizon of technology finance through technology evaluation and has been recognized for excellence by exporting the technology evaluation system to developing countries including Thailand and the European Union Commission," adding, "We will continue to strengthen Kibo’s unique role in selecting and supporting technology-based small and medium venture companies through the development of new technology evaluation models and actively respond to external demands for opening the technology evaluation model."

This content was produced with the assistance of AI translation services.