Inference of human-computation algorithms from massive-scale educational interventions

MATÍAS LÓPEZ Y ROSENFELD; MARIANO SIGMAN; DIEGO FERNÁNDEZ SLEZAK

Congreso; Doctoral Consortium - 2nd IJCAI School; 2014

The main goal of the present research statement is to develop an educational computerized framework able to detect in which tasks a child has difficulties and generate a personalized intervention based on automatic observation and evaluation of data, as part of an interdisciplinary project at the crossroads of Computer Science, Cognitive Science, Biology and Psychology. In Argentina, different government programs have provided most students with a personal laptop computer.%, for example Conectar Igualdad program with 3.000.000 laptops delivered and One Laptop per Child with 60.000 computers at La Rioja province. This unified digital platform allows educational interventions and research in a country-wide fashion. Implementing this intervention requires the development of many tools and methods. One such method constitutes the main aim of this proposal: the development of data mining techniques to infer human-computation algorithms from the huge corpus of data that is being currently collected. We aim to combine ideas from intelligent tutor systems developed at Carnegie Mellon Universitycite{koedinger2006cognitive} with training cognitive bricks (such as executive functions) at initial levels in primary school using specific games. We will focus on children in their first years of schooling (5-8 years old). By the creation of training games and their deployment country-wide, we propose the formation of a massive-scale repository of human-development cognitive data, and the methods for their analysis.