Deliverable 4.4 - INSEAD CALT

AtGentive IST-4-027529-STP - Attentive Agents for Collaborative Learners1. Introduction1.1 General introduction to AtGentiveAttention represents one of the key factors of learning performance. The most effectivelearners are not necessary the most intelligent or the brightest ones, but those who areable to (1) organise efficiently their time; (2) sustain concentrating on their key activitiesand to complete them, and (3) have the psychological strength to mobilise all theirenergy for the last miles that will really make a difference.This situation is aggravated in an online setting, where learners are left on their own,have fewer points of reference to situate themselves, do not receive any direct pressurefrom a tutor or from their peers, and can more easily procrastinate or engage in learningactivities that are very ineffective.The objective of this project is to investigate the use of artificial agents for supporting themanagement of the attention of young or adult learners in the context of individual andcollaborative learning environments. This project comprises the modification of existinglearning-support software to incorporate attention-enhancing features identified withinAtGentive’s conceptual framework (see deliverable D1.3 – “AtGentive conceptualframework and application scenarios”) and found to be desirable by the formativeevaluation. Such features range from relatively simple enhancements to facilitateperception of more relevant information, to direct intervention with the user by embodiedagents. Overall, the aim has been to enhance the learners’ effectiveness by directingtheir attention in more appropriate directions. This was approached in three broad ways:implicitly, by direct intervention, and by interventions to proactively coach the learners inthe management of their own attention (assessment, guidance, stimulation, etc.).Interventions are controlled by agents that profile the (short or long term) state of theattention of the learners by observing their actions, to assess, to analyse and to reasonon these states of attention and to intervene as suggested by the conceptual framework.Where agents need to communicate directly with the learners, they may do so simply bychanging the information on the screen, adjusting what is available to the user. Moredirectly, they are able to appear as cartoon-style characters, embedded in the applicationand its interface. Thus, embodied agents are an important interface element forAtGentive.Interventions have been designed and tested as part of two different learninginfrastructures / contexts. The initial focus for one context, AtGentSchool, is selectedschools in the Czech Republic. It supports students aged between 8 and 15 years of agecollaborating with subject experts. AtGentSchool is built on the Ontdeknet eLearningplatform, created by the Dutch company Ontdeknet. This platform is an electroniclearning environment that makes knowledge and skills in society accessible toeducational institutions in general and individual students in particular. Virtual learningrelationships between subject experts and students are established in this virtual learningenvironment. The Ontdeknet environment provides guidance to support individuals tolearn together based upon common interests. This platform originally used an embodiedagent (“Onty”, a cartoon fish) to guide learners around the learning environment. Thesoftware was adapted to incorporate the AtGentive interventions, and the existingembodied agent changed in accordance with the results of formative evaluation to“Honza” (a cartoon boy) (see Figure 1 and deliverable D4.2 – “Result of the FormativeEvaluation”.The second context, AtGentNet, focuses initially on use by adult learners enrolled inbusiness-related courses organised by the Swedish Trade Council (STC). It will supportDel 4.4: AtGentive Final Evaluation Report page 1