Adaptive PE-HRI: Data for research on Social Educational Robots driven by a Productive Engagement Framework

This dataset corresponds to our work on developing autonomous social educational robots (namely Harry and Hermione) driven by a productive engagement framework in open ended collaborative learning environments. The data is collected in the context of a robot mediated collaborative and constructivist learning activity called JUSThink where each team interacts with the activity for around 1 hour consisting of a 30 minute collaborative play. 

In this data set, team level multi-modal behavioral data is collected from 52 teams of two (104 children) where the children are aged between 9 and 12. The definitions are given below: 

condition: This column indicates which condition do the teams belong in. 0 and 1 for teams interacting with Harry and Hermione, respectively.

Error: This is the error of the last submitted solution. Note that if a team has found an optimal solution (error = 0) the game stops, therefore making last error = 0. This is a metric for performance in the task. 

Learning Gain: It is a team-level learning outcome defined as the difference between the number of questions that both of the team members answer correctly in the post-test and in the pre-test, which grasps the amount of knowledge acquired together by the team members during the activity.

Usefulness Score: The score quantifies the team's subjective evaluation of a robot intervention in terms of it's usefulness as perceived by each team member individually. The score can assume values of 1, 0, 0.5 if both found the suggestion useful, not useful, or if they differed in their evaluation, respectively

PE Score: It is a quantification of the Productive Engagement state of the team, computed on the basis of quantifiable observable behaviors found conducive to learning in training phase

Right_Suggestions: This metric captures the team's subjective evaluation of the robot's competence on a five-points likert scale to the statement "I think the robot was giving us the right suggestions". It is an average of the team member's individual answers. 

Right_Time: This metric captures the team's subjective evaluation of the robot's competence on a five-points likert scale to the statement "I think the robot gave us suggestions at the right time". It is an average of the team member's individual answers.

Exploration: This variable represents how many interventions of Exploration type were received by a particular team normalized with respect to the entire data set. 

Reflection: This variable represents how many interventions of Reflection type were received by a particular team normalized with respect to the entire data set. 

Communication: This variable represents how many interventions of Communication type were received by a particular team normalized with respect to the entire data set. 

LG_status: This column indicates if a team belongs to a high learning or low learning group based on a mean split on the entire data set. 

This dataset corresponds to the publication "Social robots as skilled ignorant peers for supporting learning": https://doi.org/10.3389/frobt.2024.1385780