Transcript of "The meaning behind smileys - presentation at EMPIRE 2013 workshop at UMAP2013"
The Meaning Behind Smileys – AnAffect Self Report Tool Based OnEmpirical DataAdam Moore (@adam__moore),Christina M. Steiner& Owen Conlan
• How to measure Affect?• Self report avatars• SBAI• Online survey• Pilot• Main Results• Questions . . .Overview
• Facial / Behavioral Analysis• Special equipment• Interferes with learning experience?• Data intense• Textual Analysis• Quite a lot of text affect neutral• Requires good models of affect expression• Self-reports• Require an internal awareness• Interrupts flow & fidelity?How to measure Affect?
Why not just use smileys?SmileyBasedAffectIndicatorA RESTful affect self-report toolbased on empirical data
SurveyDemographic Info:• Gender• Age• Country of birth• Residence• Education• Internet usage
Adjustment after pilotSmiley 5 & 6 too similar, so 6 changed tobe more neutral
• 996 complete replies were received in 2 months.• The cohort was composed of 285 women, 700 men and 11 respondentspreferred not to say.• Reported ages ranged from 15 to 103, with an average of 26.7 (SD 10.4).• Analytics point to a large number of responses to have been made inanswer to the mailing to Trinity College; so cultural referents are skewedas a result. For example, nearly 70% of respondents give their country ofbirth as Ireland, and over 90% give Ireland as their country of currentresidence.Online survey
Smiley valence vs arousalArousal/ActivityValence / Magnitude
• Why these smileys?• Didn’t have the one they wanted• Graphics too much – why not text?• One word is not possible• Context . . .Comments
Current Usage• Recently used in online learning simulation• Optional – displayed alongside feedback• Not much usage – 152 entries over 6 weeks• Feedback:• Not sure what it is for• Why do you need to know?• How will it effect my work / score?
• What did we do with the input?• Supports metacognitive scaffolding• Rule based – new rules on affect state• Prompts categorized to be encouraging, neutral,• Affect Text added . . .• Next look at timing / interruptionsCurrent Usage
• Much better statistics!• Analysis based on sense words• nGrams• Sense distance - wordnet• Ekman’s basic emotions• Interface refinement• Offer sense words from stemmed list?• Reflection – Mirror MoodMapApp?• Personalization / tuningStill to do . . .
• Mapping of data to cohort survey• User trial had full characterization survey• Demographics• Swedish Survey of Personality• Learning Styles (but see !!!)• Metacognitive Awareness Inventory • Social Media Attitudes (UMAP late breaking )• Look at correlations• Stereotype construction . . . Brown, E. J., Brailsford, T. J., Fisher, T., Ashman, H. L., & Moore, A. (2006). Reappraising cognitive styles in adaptive web applications. Proceedings ofthe 15th international conference on World Wide Web - WWW ’06 (p. 327). New York, New York, USA: ACM Press. Schraw, G., & Sperling Dennison, R. (1994). Assessing metacognitive awareness. Contemporary Educational Psychology, 19(4), 460–475. Adam Moore, Gudrun Wesiak, Christina M. Steiner, Claudia Hauff, Declan Dagger, Gary Donohoe, Owen Conlan (2013) Utilizing Social Networks forUser Model Priming: User Attitudes UMAP2013 Late Breaking ResultsStill to do . . .
• The research leading to these results has received funding from theEuropean Communitys Seventh Framework Program (FP7/2007-2013)under grant agreement no 257831 (ImREAL project) and could not berealized without the close collaboration between all ImREAL partners.Acknowledgements