The web-site is now in readonly mode. Login and registration are disabled. (28 June 2019)

Towards discourse-centric learning analytics

Cloud created by:

Rebecca Ferguson
8 April 2011

Presentation by Simon Buckingham Shum (co-author Anna De Liddo) as part of the learning analytics sympsium at CAL 2011.

A key indicator of meaningful learning is the quality of contribution to discourse, which is often in the written modality in online environments. Nonetheless, research on online deliberation and argumentation suggests that a more visual and hypertextual way of representing online discourse can better support learners in reflecting and making sense of what they learn while engaged in dialogue.

Several online deliberation and argumentation tools have been proposed in the literature (van den Braak et al.), which target different audiences, such as planners and policymakers, enterprise knowledge workers, online users and citizens in general. Between these, we propose Cohere, a social Web tool designed to target lifelong online learners.

Cohere aims to engage learners in structured online discourse, and to support a move from unstructured textual dialogue to structured and visual ideas sharing and connections.

In this paper we discuss what it means to use Cohere to monitor online learning activities and develop useful learning analytics, by starting with analysis of the online discourse in which learners are involved.

In particular, we propose to develop what we define as discourse-centric analytics, which are semi-automated learning analytics based on the rhetorical role that a contribution is making to a document, or conversation (e.g. identifying a problem, responding to a query, challenging or supporting a viewpoint, contributing new data). We argue that discourse-centric learning analytics enable more in-depth reflections on learners’ activities, and can augment the quality of the inferences that can be made about where and how learning happens.

To achieve this, we first present Cohere as a tool to support learners to:

• collaboratively annotate Web resources with their ideas,

• classify these ideas according to their rhetorical role, and

• connect them into a network of meaningful rhetorical moves (the social semantic network).

Moreover, we give concrete examples of how, by analysing this network, Cohere can provide learning analytics to identify:

• learners’ attention: what did learners focus on? That is: what problems and questions they raised, what comments they made, what viewpoints they expressed etc

• learners’ social interactions: direct and indirect interactions can be identified by measuring the semantic distance of learners’ contributions (ideas, comments, questions, answers etc). A social network can be drawn in which connections between learners are weighted based on the quality of the discourse in which they have been involved.

• learners’ process of meaning making: how do learners map their thoughts? Learner by learner, we can visualize the semantic-connections network they have created and compare them. This can provide analytics on how different learners, who read the same data, map their thoughts.

Based on these examples, we finally argue that discourse-centric learning analytics enable in-depth reflections on learners’ activities that would be otherwise difficult to achieve with quantitative statistical analysis of lower-level learners’ actions (such as computation of large-scale learners’ logs on how many resources they have downloaded, how much time they have spent logged into a system, how many comments they have made, how long they have spent on a task etc).

The reason for this is that discourse-centric learning analytics are based on data that makes explicit learners’ cognitive context (e.g. what kind of rhetorical move individual learners wanted to make with a comment, what meaning they gave to a connection, what contrasting viewpoints they detected etc). By analyzing more richly expressed and structured data, discourse-centered analytics can augment the level of accuracy and the cognitive depth of the inferences that can be made as to where and how learning happens.

References

van den Braak, S. W., van Oostendorp, H., Prakken, H., & Vreeswijk, G. A. W. 2006. A critical review of argument visualization tools: do users become better reasoners? In Working Notes of the 6th Workshop on Computational Models of Natural Argument (CMNA2006).

Buckingham Shum, S., 2008. Cohere: Towards web 2.0 argumentation.           In 2nd International Conference on Computational Models of Argument (COMMA 2008).

Extra content

Embedded Content

where things may be heading...

where things may be heading...

added by Simon Buckingham Shum

Contribute

Contribute to the discussion

Please log in to post a comment. Register here if you haven't signed up yet.