Analyzing Networked Learning Texts

Abstract

The term ‘network’ has many meanings from computing hardware infrastructures and the connectivity between computers that these provide to software programs that provide a platform on which individuals can form interpersonal ties, to the networks of colleagues, friends and family that make up our social worlds. What unifies these applications of the word ‘network’ is the common idea of individual nodes (computers, individuals, organizations) tied by some interaction that forms a structure greater than the sum of its parts. Interaction is essential in networked learning. Although individuals may learn by retrieving information from online archives, dictionaries and encyclopedia, it is the possibilities of interaction with others from around the globe with similar, perhaps narrowly enjoyed interests that fuels the benefits of networked learning. Thus, examining social networks – including the roles and positions of actors in a social network, their influence on others, and what exchanges support and sustain the network – is an important goal for understanding networked learning processes.

However, pursuing a social network approach raises a number of methodological issues. How do we examine and evaluate the network aspects of learning, including identifying what matters in terms of learning in the online interaction space, and how do we do this on a scale that is adequate to give more than anecdotal results? While online interaction is creating a growing legacy of texts, there are few mechanisms available for processing and analyzing such data and connecting this to performance, learning or social outcomes. Patterns of posting, interactivity, and topic threads hold information pertaining to group identity, growth and maintenance; use of words, short-hands and acronyms show the extent to which groups are embedded in disciplinary or professional norms of discourse; and interaction patterns, topics, and style provide important pointers to group purpose and conduct.

To draw interaction data from these texts, two steps are needed. First, some form of automated processing is needed to reduce the large datasets to community and conversational essentials that show the relations of importance to group members; and second, assessment of these data extractions is needed to determine the usefulness and meaning of these measures to participants. When combined, these two aspects can provide useful representations of online conversations, from statistical reports to visualizations of data and interactions, each of which can help networked learners (instructors and students) better understand the social environment in which they are participants.

This talk presents our ongoing work on a novel, automated method for extracting interaction data from threaded discussions of networked learning groups and some preliminary evaluation of the results. Using natural language processing, the proposed method reduces large text-based datasets to community and conversational essentials that show the relations of importance to group members. By studying these relations, we hope to identify what matters in terms of learning in the online interaction space and to provide useful representations of online conversations to help networked learners (instructors and students) better understand the social environment in which they are participants. To do so also requires making accurate determinations of who is talking to whom. In our written paper we discuss methodological issues associated with extracting names from networked learning texts and our procedures for enhancing network information through new techniques of name extraction. In our presentation we will demonstrate the working prototype of the software tool, and discuss findings and methodological issues associated with this kind of analysis.