Detection, Analysis and Visualisation of Detection, Analysis and - PowerPoint PPT Presentation

Detection, Analysis and Visualisation of Detection, Analysis and Visualisation of Georeferenced Emotions Georeferenced Emotions Eva Hauthal & Dirk Burghardt Eva Hauthal & Dirk Burghardt Dresden – – 2013/08/26 Dresden 2013/08/26

Motivation http://www.apfelnews.eu/2009/07/31/das ‐ iphone ‐ als ‐ mobiler ‐ reisefuhrer ‐ von ‐ polyglott/ Emotional Travel Guide Detection, Analysis and Visualisation of Georeferenced Emotions 2 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Emotions Overall emotional reaction: ∙ physiological reactions ∙ tonic posture reactions ∙ instrumental motoric reactions ∙ expressive motoric reactions ∙ expressive linguistic reactions ∙ subjective experience components Battacchi et al. (1996) Emotion und Sprache: zur Definition der Emotion und ihren Beziehungen zu kognitiven Prozessen, dem Gedächtnis und der Sprache. Lang, Frankfurt am Main. Detection, Analysis and Visualisation of Georeferenced Emotions 3 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Title Title Description Description Tags Tags Longitude Longitude Latitude Latitude Detection, Analysis and Visualisation of Georeferenced Emotions 4 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Emotional Associations thing, place, memories emotion activity, ... experiences word Detection, Analysis and Visualisation of Georeferenced Emotions 5 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Structuring Emotions arousing Valence negative positive Arousal Russell (1980) A circumplex model of affect. unarousing Journal of Personality and Social Psychology 39. Detection, Analysis and Visualisation of Georeferenced Emotions 6 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

ANEW & BAWL-R Affective Norms for English Words Berlin Affective Word List ‐ Reloaded Bradley & Lang (2010) Võ et al. (2009) Affective norms for English Words (ANEW): The Berlin Affective Word List Reloaded (BAWL ‐ Stimuli, instruction manual and affective ratings. R). Technical Report C ‐ 2, University of Florida. Behavior Research Methods 41(2). Detection, Analysis and Visualisation of Georeferenced Emotions 7 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Emotional Analysis of Dresden Detection, Analysis and Visualisation of Georeferenced Emotions 8 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Emotional Analysis of Dresden Detection, Analysis and Visualisation of Georeferenced Emotions 9 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Considerations 1 place ≠ 1 emotion  personal preferences/experiences/memories  individual vs. collective emotions  temporal aspects Detection, Analysis and Visualisation of Georeferenced Emotions 10 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

< 442 words/km 2 442 – 514 515 – 956 957 – 3627 3628 – 19765 19766 – 117288 CC BY ‐ NC by lyricsart http://www.flickr.com/photos/91821458@N00/76083614/ 1 place ≠ 1 emotion  temporal aspects CC BY ‐ SA by MalteF http://de.wikipedia.org/wiki/Dresden Detection, Analysis and Visualisation of Georeferenced Emotions 11 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

< 442 words/km 2 442 – 514 515 – 956 957 – 3627 3628 – 19765 19766 – 117288 CC BY ‐ NC by _parrish_ http://www.flickr.com/photos/_parrish_/439884019/ Detection, Analysis and Visualisation of Georeferenced Emotions 12 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

< 442 words/km 2 442 – 514 515 – 956 957 – 3627 3628 – 19765 19766 – 117288 18 сецондс CC BY ‐ NC ‐ SA by http://www.flickr.com/photos/18seconds/2949364022/ Detection, Analysis and Visualisation of Georeferenced Emotions 13 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

< 442 words/km 2  temporal aspects 442 – 514 515 – 956 957 – 3627 3628 – 19765 19766 – 117288 realname ⇋ CC BY ‐ NC ‐ SA by tonal decay http://www.flickr.com/photos/19363084@N07/5459318526 Detection, Analysis and Visualisation of Georeferenced Emotions 14 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Grammatical Special Cases happy based on Biedermann (1969) Die deutschen Gradadverbien in synchronischer und diasynchronischer Sicht. Dissertation, Heidelberg. Quirk et al. (1985) A comprehensive grammar of the English language. Longman, London. van Os (1989) Aspekte der Intensivierung im Deutschen. Narr, Tübingen. Detection, Analysis and Visualisation of Georeferenced Emotions 15 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Grammatical Special Cases  degree words modifying the intensity of... adjectives It was absolutely good . adverbs They love their dog very dearly . verbs He denied it strongly. nouns It was really an adventure . based on Quirk et al. (1985) A comprehensive grammar of the English language. Longman, London. Detection, Analysis and Visualisation of Georeferenced Emotions 16 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Modification of Valence and Arousal Values - Negation e.g. unhappy  happy V O = 2.4075 A O = 3.745  unhappy V neg = ‐ 2.4075 A neg = 2.255 V neg = ‐ V O A neg = 6 ‐ A O Detection, Analysis and Visualisation of Georeferenced Emotions 17 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Detection, Analysis and Visualisation of Georeferenced Emotions 18 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Detection, Analysis and Visualisation of Georeferenced Emotions 19 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

„... totally „... entirely destroyed...“ destroyed...“ „... badly damaged ...“ CC BY ‐ SA by Deutsche Fotothek http://www.deutschefotothek.de/obj88880671.html „... completely destroyed...“ CC BY ‐ SA by Deutsche Fotothek http://www.deutschefotothek.de/obj88950461.html Detection, Analysis and Visualisation of Georeferenced Emotions 20 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Future Work ∙ time dependence & temporal aspects ∙ distinction individual / collective emotions ∙ visualisation Mobile Emotional Travel Guide ∙ adaptation to ‚emotional needs‘ of user ∙ emotional routing ∙ emotional profiles of places/sights Detection, Analysis and Visualisation of Georeferenced Emotions 21 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Thanks for your attention! Thanks for your attention! eva.hauthal@tu eva.hauthal@tu ‐ ‐ dresden.de dresden.de dirk.burghardt@tu ‐ ‐ dresden.de dresden.de dirk.burghardt@tu

Extracting Emotions from Language Zarafani et al. (2010) Abbasi et al. (2008) Sentiment Propagation in Social Networks: Affect Analysis of Web Forums and Blogs A Case Study in LiveJournal. Using Correlation Ensembles. Proceedings of Advances in social IEEE Transactions on Knowledge and Data computing, Third International Conference Engineering 20(9). on Social Computing, Behavioral Modeling, and Prediction. Detection, Analysis and Visualisation of Georeferenced Emotions 23 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Extraction of Emotions from Photo-Metadata Title Description Tags Sentence Segmentation Language Detection POS Tagging Lemmatisation Word Valence Arousal Longitude ANEW / Latitude BAWL ‐ R yes Word Valence yes no ANEW / Found Hypernym / Found Arousal BAWL ‐ R Word? Word? Synonym Longitude Latitude Detection, Analysis and Visualisation of Georeferenced Emotions 24 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Emotional Analysis of Dresden with Consideration of GSC Detection, Analysis and Visualisation of Georeferenced Emotions 25 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Error Analysis in 50 Test Data Records Detection, Analysis and Visualisation of Georeferenced Emotions 26 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Modification of Valence and Arousal Values - Negation 5 4 'happy' (ANEW) Arousal 3 'unhappy' (formula) 'unhappy' (ANEW) 2 1 ‐ 3 ‐ 2 ‐ 1 0 1 2 3 Valence Detection, Analysis and Visualisation of Georeferenced Emotions 27 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Modification of Valence and Arousal Values Maximum Valence Arousal Amplification Valence Arousal Attenutation Valence Arousal Minimum Valence Arousal Detection, Analysis and Visualisation of Georeferenced Emotions 28 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Detection, Analysis and Visualisation of Georeferenced Emotions 29 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -

Modification of Valence and Arousal Values - Amplification e.g. very happy  happy V O = 2.4075 A O = 3.745  very happy V amp = 2.883 A amp = 4.2125 2 ‐ |A O – 3| 3 ‐ |V O | 3 3 3 ‐ |V O | 2 ‐ |A O ‐ 3| ∙ +1 ∙ +1 3 3 |V O | A amp = 2A O ‐ 3 ‐ ½ ∙ |(A O ‐ 3)| ∙ (A O ‐ 3) V amp = (2 ‐ ) ∙ V O 3 Detection, Analysis and Visualisation of Georeferenced Emotions 30 Dresden – 2013/ 08/ 26 - Eva Hauthal & Dirk Burghardt -