Lukewarm Starts for Computerized Adaptive Testing Based on Clustering and IRT.

In the field of educational data, the usage of recommendations can considerably improve the experience of the students during the training or assessment phases. In such cases, the activities have to be well adjusted to the latent knowledge of specific students groups in order to individualize, alleviate and speed up the process. We propose a framework to avoid cold starts by recommending initial items inputting a classical Computerized Adaptive Testing. Our global approach is supported by the hybrid usage of Item Response Theory and techniques of clustering to build tailored evaluation paths. To illustrate our methodology, we use real data from the Brazilian National High School Exam.