Measures of group knowledge structure convergence in an online collaborative learning environment

RClariana@psu.edu (Penn State)

• Post lesson essays, measure of group knowledge convergence, collaboraitve onlien learning community of practice
• decomposed to data arrays using ALA-Reader, analyzed using Pathfinder network analyssis
• degree centrality and graph centrality to find knowledge convergence
• participants: Subaru Dealership shop coordinators
  • isolated in their workplaces, real CoP
• Dimensional semantic space
  • when you do something to individual, they become more alike
• Study
  • self-paced lesson (2 months)
  • one group with CoP, one without
  • post-lesson essay
  • essay prompt is critical
• Analysis approach
  • essay → (ALA Reader) → prx data array → (KNOT) → PFNET
  • important terms from expert answer, picked automatically
  • force-directed map, CMAP tools
  • we want many individual's averaged into one
    • many essays → many arrays → one average data array → one PFNET
    • split into two random groups to see variability, to four groups (CoP 1 + CoP2 + Control1 + Control2)
    • the expert's node centrality, table of node centrality
    • node centrality → graph centrality
      • linear
      • hierarchical
      • network
      • star
    • similarity of graphs - what he calls “convergence”
  • on first approach, identify most commonly used words for each concept, then normalize (car, vehicle → automobile etc)