-
Learning is a complex, mostly invisible process that happens across spaces, occurring
in the physical world but also increasingly in virtual worlds or web-based spaces. In
order to explore what happens in such blended learning experience, there is a need for
multiple data sources that bring evidence from these different spaces, including logs,
learning resources, or even physical sensors [
1
]. The combination of different data
sources often generates multimodal datasets, with data representing different views of
the same learning event. Moreover, multimodal analyses can be applied, contributing
to a richer, triangulated view of the learning process [
2
].
Two workshops co-located with LAK‘17 in Vancouver (Canada) -the 2nd Cross-LAK
and the 6th Multimodal Learning Analytics (MMLA) workshop- tackle the analysis of
this complexity, from slightly different perspectives. While the former focuses on the
challenges imposed by the multiplicity of learning contexts (i.e., spaces), the latter
explores different data sources and solutions that may help address those challenges.
The goal of the 2nd Cross-LAK workshop was to gather the sub-community of LAK
researchers, learning scientists and researchers from other communities (e.g. artificial
intelligence in education, educational data mining, intelligent tutoring systems, etc),
interested in ubiquitous, mobile and/or face-to-face learning analytics. An overarching
concern tackled by this series of workshops is how to integrate and coordinate
learning analytics to provide continued personalised support to learning across digital
and physical spaces, i.e. considering the ecologies of devices and learning (and non
learning tools) that are used in real-world contexts. The particular goal of this second
workshop was to define the 5-year vision of learning analytics. Participants were
prompted to move away from the assumptions and constraints commonly imposed by
current learning analytics solutions (e.g. which often focus on clickstreams to model
student’s behaviours) to embrace a more holistic view of learning, which occurs a
different spaces, both physical and digital. The workshop received 7 participant
submissions, 5 of which were accepted. A total of 25 people attended the workshop.
In turn, the MMLA sub-community has had as its overarching goal to explore and
understand how to go beyond the current state of learning analytics that derives
insights from only clickstream or other single data source [
1
]. The aim of this
particular MMLA workshop was twofold: to chart the landscape of this research area
and collaboratively identify a set of grand challenges to be addressed by the MMLA
community. The workshop received 12 submissions, all of which were accepted.
Eleven of them are included in the present volume. A total of 21 people attended the
workshop. Inspired by the proposals submitted to the workshop, during the event the
participants ideated MMLA solutions devoted to support key problems of different
learning contexts (individual and collaborative learning in primary-secondary school,
university and workplace settings). From the discussion about these solutions and the
participants’ own experience, MMLA challenges were identified regarding the data
gathering (e.g., synchronization, lack of standards for data representation, integration
of third-party data, and physical intrusiveness), the data analyses (conceptual and
technical integration problems, extraction of high level indicators from raw data, and
later modelling, training and segmentation), the feedback to the stakeholders (data
literacy, actionability of MMLA across stakeholders), and the adoption (privacy,
ethical, cultural challenges, sustainability, and data literacy).
There exist several commonalities and differences between the multimodal (MMLA)
and across-spaces (Cross-LAK) approaches to learning analytics. The across-spaces
approach highlights scenarios where student’s learning activities are not constrained
to a single physical or digital environment [
3
]. Students may interact face-to-face or
remotely, and use diverse (educational) tools, depending on the context, subject
matter or the informal learning opportunities. Pervasive and mobile technologies can
be used to allow learners to get remote access to educational resources from different
physical spaces (e.g. ubiquitous/mobile learning support) or to enrich their learning
experiences in the classroom in ways that were not previously possible (e.g.
face-to-face/blended learning support). The MMLA approach is focused on capturing,
integrating and analysing learning traces from different sources, even if the event
occurs in just one space, in order to obtain a deeper understanding of the learning
process, wherever it happens. In this case, the spotlight is on the diverse sensors,
different frequencies of data collection, and the sophisticated machine learning and
artificial intelligence techniques that can be used to interpret such complex data. So
far, the majority of MMLA studies have been conducted under semi-controlled
conditions [
4
].
Furthermore, Cross-LAK approaches often take a top-down approach, in which the
first step is to understand the complexity of the learning situation in its totality (which
commonly occurs in the wild). Only then the approaches would consider the
application of means for capturing data and applying analytics techniques to discover
insights (often including multimodal analytics techniques). In contrast, MMLA
approaches tend to take a bottom-up perspective that starts from multimodal data and
tools. Then, based on the insights from the analysis of these data, the aim is to provide
a more complete view of the associated learning and teaching situation.
The following joint proceedings volume aims to promote the dialogue and the
alignment of research efforts across both sub-communities. Moreover, this
collaboration is the seed of a Special Interest Group (SIG) that will be part of the
Society of Learning Analytics Research (SoLAR)1. The goal of this SIG will be to
advance the understanding of the learning process, no matter where and how it
happens.