This study focuses on the design of several social agents that are intended to collect the self-reflections of learners while learners are immersed in simulation activities for knowledge building. The design of the agents follows 5W principles and seeks to encourage learners to expend mental effort upon multi-faceted learning and self-reflection. Using semantic networks, we developed dialogue lines for reflection-prompting agents. We analyzed the participants' answers using natural language processing technology to classify the sentences into positive and negative rankings. A preliminary field study with 117 high school students was conducted over three weeks to test the effects of agent-prompted self-reflection. The results demonstrated that 96% and 62% of participants separately completed the first and the second simulation activities (including the agent-prompted self-reflections respectively). Those who did not finish the activities were generally limited by time restrictions rather than a lack of motivation, as the participants typically considered the interactions with the agents to be interesting. The self-reflections elicited through the agent interviews were consistent with the reflections obtained from paper-pencil questionnaires and appeared to be stable over time. Future study, including investigations using a randomized experimental design with a control group, is needed to fully assess the effects of agent-prompted self-reflection.
|Number of pages||15|
|Journal||Educational Technology and Society|
|State||Published - Jul 2013|
- Intelligent agent
- Multiple intelligence