Riiid, on Friday, announced that it acquired a patent for enhanced learning algorithms and framework technologies. The edtech company’s technology recommends optimal problems to improve learner’s score prediction accuracy.
The patent involves a dedicated “expert model” operation method that generates a score prediction model using big data and artificial intelligence. The software collects large amounts of learner problem-solving data to train the model efficiently. It then recommends a diagnostic test problem based on the existing score prediction model.
The expert model performs all functions such as problem-solving, data analysis, diagnostic examination, problem selection, and recommendation. It aims to improve the accuracy of the score prediction model learning and to advance itself.
In particular, the reinforcement learning algorithm performs millions of problem-solving simulations to questions not solved by the student. It then predicts and analyzes the results of the students’ solutions, making it possible to forecast a more efficient and accurate score prediction.
The new software overcomes the effective limitations of the existing score prediction model, which predicted scores by calculating the correct answer rate for every problem that the student would solve in the future.
With the patent acquisition, Riiid secured AI technology capable of predicting and diagnosing high-accuracy scores with minimal multiple-choice questions in environments that require a variety of learning and condition diagnosis. The AI TOEIC tutor “Santa,” which Riiid uses, predicts the user’s score with 90% accuracy with a diagnostic test of 6 to 10 problems.
Technological Expansion Through Patents
Riiid already acquired a total of 15 domestic and foreign patents, including the latest obtained patent. The company currently has 70 more patents pending.
The South Korean company presently fast-tracks the assetization of services across all areas of AI tutoring. These include services such as the composition of the diagnostic test problem to the optimal problem recommendation, score and session dropout prediction, problem similarity calculation, AI modeling architecture, and AI products technology for optimized UI and user interaction.
Riiid said it is also focusing on securing technology rights in major global markets such as the U.S., Japan, China, and Vietnam through patent applications in Korea as well as overseas.