An Architecture for a Decentralised Learning Analytics Platform (Positioning Paper)

Ekuban, Audrey and Domingue, John (2023). An Architecture for a Decentralised Learning Analytics Platform (Positioning Paper). In: CEUR Workshop Proceedings: SEMMES 2023: Semantic Methods for Events and Stories workshop ESWC 2023 (Alam, Mehwish; Trojahn, Cassia; Hertling, Sven; Pesquita, Catia; Aebeloe, Christian; Aras, Hidir; Azzam, Amr; Cano, Juan; Domingue, John; Gottschalk, Simon; Hartig, Olaf; Hose, Katja; Kirrane, Sabrina; Lisena, Pasquale; Osborne, Francesco; Rohde, Philipp; Steels, Luc; Taelman, Ruben; Third, Aisling; Tiddi, Ilaria and Türker, Rima eds.), CEUR Workshop Proceedings (CEUR-WS.org), 3443.

URL: https://ceur-ws.org/Vol-3443/ESWC_2023_TrusDeKW_pa...

Abstract

Predictive Learning Analytics is a subfield of Learning Analytics that helps identify students who are at risk of dropping out or failing. However, the centralised approach to Predictive Learning Analytics raises privacy and ethical concerns, particularly in the area of data collection. In this paper emphasis is placed on a decentralised mechanism for collecting student consent. This mechanism is part of the EMPRESS framework, that combines self-sovereign data, Federated Learning, and Graph Convolutional Networks for Heterogeneous graphs to address these issues. EMPRESS allows data owners to control who has access to their data, processes data on their devices, and utilizes knowledge graphs for analysis. Course tutors can use the insights provided by the analyses to offer timely assistance to these students. In addition, this paper details how Heterogeneous graphs can be used in Predictive Learning Analytics.

Viewing alternatives

Item Actions

Export

You can export this page using these formats