Resumo

SALAZAR GOMEZ, Lisset; VAZQUEZ SANCHEZ, Angel Alberto  e  CANIZARES GONZALEZ, Roxana. Estimating latent knowledge in large volumes of data using the Bayesian Knowledge Tracing algorithm. Rev cuba cienc informat [online]. 2021, vol.15, n.4, suppl.1, pp. 165-180.  Epub 01-Dez-2021. ISSN 2227-1899.

Given the massive amount of data generated in education, the traditional methods for knowledge discovery have had to be changed. One of the algorithms is the Bayesian Knowledge Tracing (BKT) that allows to Estimate Latent Knowledge (LKE). LKE is nothing more than a way of measuring a student's knowledge about specific skills and concepts, which is evaluated by his or her correction patterns on those skills. This algorithm is designed to be used on small volumes of data, affecting its performance in the presence of large volumes of data. In order to solve the problem, the transformation of the BKT algorithm will be presented as a result, taking into account parallel and distributed programming. Parallel processing tools such as the Apache Spark framework were used in a mining environment. The proposed solution is validated through tests to measure performance and effectiveness, using metrics such as speedup, efficiency, mean squared error of the differential of probabilities and mean squared error of the differential of the area under the ROC curve; educational databases were used for the tests.

Palavras-chave : Latent Knowledge Estimation (LKE); Educational Data Mining (EDM); Bayesian Knowledge Tracking (BKT).

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )