1 / 22

Conversational Strategies that Support Idea Generation Productivity in Groups

Conversational Strategies that Support Idea Generation Productivity in Groups. Rohit Kumar Jack L. Beuth Carolyn P. Rosé. Overview. User Study: Design of Tutor prompts for supporting Collaborative Learning Software Agents: Monitoring Student Interactions Social Interaction Strategies

karis
Download Presentation

Conversational Strategies that Support Idea Generation Productivity in Groups

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Conversational Strategiesthat SupportIdea Generation Productivityin Groups Rohit KumarJack L. BeuthCarolyn P. Rosé

  2. Overview • User Study: • Design of Tutor prompts for supporting Collaborative Learning • Software Agents: Monitoring Student Interactions • Social Interaction Strategies • Presentation Style • Outline: • Background: Research Objectives, Earlier Work • Presentation Style:Heteroglossia • User Study: Task, Procedure, Metrics, Design • Results Background  Presentation Style  User Study  Idea Generation  Results

  3. Bridges CollaborativeLearning TutorialDialog DialogSystems Our Work CommunicationStudies Small GroupCommunication ConversationalTutors Background Presentation Style  User Study  Idea Generation  Results

  4. Prior Work • Benefits of • 0.90σ : Collaborative Learning • 1.06σ : Dynamic Support (Software Agents) • 1.24σ : Combined • Kumar et. al., 2007 • Students ignore/abuse Tutors • Tutors are unable to manage theattention of the students Background Presentation Style  User Study  Idea Generation  Results

  5. Social Interaction Strategies • Small Group Communication • Instrumental vs. Socio-Emotional Interaction • Bales, 1950 • Kumar & Rosé, 2010 Background Presentation Style  User Study  Idea Generation  Results

  6. Social Interaction Strategies • Implemented Using Basilica • Kumar & Rosé, 2011 (IEEE Trans. Learning Tech.) Background Presentation Style  User Study  Idea Generation  Results

  7. Social Interaction Strategies • Significant Benefits • 0.71σ : Using Fully-Automated Social Interaction Strategies • Kumar et. al., 2010 (ITS) • More Efficient at delivering the instructional content • Other Experiments • Amount of Social Behavior • Learning, On-Task Interaction Hua et. al., 2010 (ITS) • Why/How & Timing of Behavior • Learning, Perception of Tutor Kumar & Rosé, 2011(SigDial) • Transfer to other domains • Current Work • Style of Presentation • This talk Background Presentation Style  User Study  Idea Generation  Results

  8. Presentation Style • Heteroglossia vs. Monoglossia • Lexical/Grammatical choices • Indicates social positioning of the speaker • Heteroglossia • Awareness/Acknowledgement of multiple perspectives Background Presentation Style  User Study  Idea Generation  Results

  9. User Study • Design • 2 x 3 • Full factorial • Between subjects • Using Social Interaction Strategies • Yes (S) • No (NS) • Style of Presentation • For Tutor Prompts • Original (N) • Monoglossic (M) • Heteroglossic (H) Background  Presentation Style User Study  Idea Generation  Results

  10. User Study: Learning Activity • Mechanical Engineering Course • Freshmen: Wrench Design • Underlying Concepts • Force, Moment, Stress, Strength, … • Collaborative Design Lab • Design a better wrench • Ease of Use, Cost, Safety • Team of 3-4 students • Communicate using ConcertChat • Supported by Automated tutors • Randomly assigned to conditions Background  Presentation Style User Study  Idea Generation  Results

  11. User Study: Procedures • Pre-Test: 10 minutes • Qs: 7 multiple-choice, 4 brief explanation • Collaborative Design Competition: 35 minutes • Interact with tutors • Post-Test: 10 minutes • Perception Survey: 5 minutes • Details in the paper • 10 items 7-point Likert scale • Personality descriptors Background  Presentation Style User Study  Idea Generation  Results

  12. User Study • Metrics • Learning Outcomes • Perception Ratings (of Tutor , Task) • Idea Contributions (by Students) • Count is estimated using dictionary • how about, can we, we should, … • increase, decrease, steel, … • Noisy Measure • For all conditions Background  Presentation Style User Study  Idea Generation  Results

  13. Idea Generation: Review • Idea Generation & Learning • Wang et. al., 2007 (Upcoming IJCSCL Article) • Depends on the Task • Idea Generation in Groups • Process Loss • Affect Productivity  Reduce Learning • Objective: • To create support that enable the benefits of idea generating interactions with negative side-effects • Instrument: • Automatically generated intervention prompts Background  Presentation Style  User Study Idea Generation  Results

  14. Results • Learning Outcomes: • No significant effects of either factor • Perception Ratings: • Heteroglossic prompts: • Happier with the discussion p<0.05 • As good as human tutor p=0.08 • Perceived task success p<0.08 • Heteroglossic & Monoglossic prompts (i.e. Style Consistency) • Liked the tutor very much p<0.05 • With Social Interaction Strategies • Tutor perceived more supportive & less pushy p<0.05 Background  Presentation Style  User Study  Idea Generation Results

  15. Results • Idea Generation Productivity • As estimated by counts of idea contributions • Social Interaction Strategies • No Significant Effect • Presentation Style • Heteroglossic > Original • Heteroglossic > Monoglossic • Interaction Effect • p<0.05 • Heteroglossic prompts could compensate for lack of social capabilities Background  Presentation Style  User Study  Idea Generation Results

  16. Conclusions • Exploration of style / function of tutor prompts • Among other factors (amount, timing, transfer, …) • Perception Ratings • Preference towards Heteroglossic prompts & the use of Social behavior • Idea Generation • Heteroglossic style encourages more productive idea generation • Observation: • Unlike Wang et. al. 2007, no relationship between idea generation & learning

  17. Thank You • Acknowledgements • Colleagues • Gregory Dyke • CMU Mechanical Engineering Faculty & TAs • Maarten P. De Boer, Philip Leduc, Robert Reid, Joshua Eickmeier, Mallory Elbert, Lina Gonzalez, Prerna Singh • C-MITES • Ann Shoplik, Pamela Piskurich • National Science Foundation Questions

  18. Prior Work • Benefits of • 0.90σ : Collaborative Learning • 1.06σ : Adaptive Support (Tutor Agents) • 1.24σ : Combined • Kumar et. al., 2007 • Students ignore/abuse Tutors • Remedial Interaction Tactics • Attention Grabbing • Chaudhuri et. al., 2008/2009 Background Presentation Style  User Study  Idea Generation  Results

  19. Presentation Style • Heteroglossia vs. Monoglossia • Lexical/Grammatical choices • Indicates social positioning of the speaker • Heteroglossia • Awareness/Acknowledgement of multiple perspectives • Criteria • Martin & White, 2005 • Propositional content must be asserted • Awareness to alternative perspective must be visible • Awareness is expressed grammatically or paraphrastically Background Presentation Style  User Study  Idea Generation  Results

  20. Presentation Style • Examples Background Presentation Style  User Study  Idea Generation  Results

  21. Presentation Style • System of Engagement • Martin & White, 2005 • Limited exploration of this system in our study Background Presentation Style  User Study  Idea Generation  Results

More Related