Читать книгу Rethinking Digital Native Comunicators Training - Cristina Pulido Rodriguez - Страница 13

In 2020, the team in UCLouvain tested a new kind of teaching experiment. The objective was to immerse students in the effects and functioning of recommendation algorithms via a secure simulation. Without the technical limitations we faced on Facebook and with the possibility to control all the parameters.

Every week, for 12 weeks, students connect anonymously to a closed and secured news website, presenting them with 15 articles from Belgian reference news websites. Students liked, disliked or shared articles according to their usual tastes and preferences. Sharing was done within the group. The whole collection of “traces” was anonymized, stored and analysed every week. The teaching staff showed them the “reports” each week: most liked and shared articles, favourite sections, etc.

At the end of the 12 weeks, the students had to take back all the group reports and trends and do a self-analysis of their profile. This allows each person to measure the gap between their true profile and their perception of themselves, and to measure the degree of similarity with the group trends.

Finally, our engineering colleagues trained a recommendation algorithm with the data from the experiment and were supposed to submit recommendations to each student anonymously, so that they reflect critically on what they ‘receive’ as recommendations, and what effects this has on them. Unfortunately, the feedback session that was planned during the spring of 2020 had to be cancelled due to health restrictions, so they did not submit these recommendations to the students (it seemed inappropriate to “throw” it at them without feedback) but this was the planned scenario.