Editorial for the 3rd AAAI-20 Workshop on A ective Content Analysis

Abstract

A Con2020, the third AAAI Workshop on A ective Content Analysis @ AAAI-20, focused on interactive a ective content, i.e., analysis of emotions, sentiments, and attitudes in textual, visual, and multimodal content for applications in psychology, consumer behavior, language understanding, and computer vision especially in conversational content. It included the second CL-A Shared Task on modeling self-disclosures. The program comprised keynotes, original research presentations, a poster session, and presentations by the Shared Task winners.

Introduction

Workshop Topics and Format

Presentations at the workshop featured psycholinguists, computer science researchers, and experts in marketing science. Topics included new approaches that address open problems such as deep learning for a ect analysis, leveraging traditional a ective computing (multimodal datasets), privacy concerns in a ect analysis, and inter-relationships between various a ect dimensions. These fall under the broad topics of interest of the workshop: { Deep learning-based models for a ect modeling in content (image, audio, and video) { Psycho-demographic pro ling { A ective and Cognitive Content Measurement in Text { A ect in communication { A ectively responsive interfaces { A ective human-agent, -computer, and -robot interaction { Mirroring a ect { A ect-aware text generation { Measurement and evaluation of a ective content { Consumer psychology at scale from big data { Modeling consumer's a ective reactions { A ect lexica for online marketing communication { A ective commonsense reasoning { Multimodal emotion recognition and sentiment analysis { Computational models for consumer behavior { Psycho-linguistics, including stylometrics and typography { Computational linguistics for consumer psychology 3

Overview of the papers

The workshop featured ve keynote talks, three paper presentations, and two poster sessions. 38 papers were submitted to the workshop, 6 of which were Systems for the CL-A shared task. Finally, 4 papers were accepted as full papers, and 4 were accepted as posters, and these will be included in the proceedings. In addition, the winners from the CL-A task presented talks and posters at the workshop. One pre-published paper was also invited for an invited talk.

The following sections brie y describe the keynote and sessions. 3.1

Keynotes

The rst keynote by Prof. Louis-Philippe Morency from CMU was about Multimodal AI, speci cally around understanding human-computer interactions and dynamics. The talk started with laying down a foundation around human-agent (computer) interactions and the role of a ective interactions in that setup. Further, he discussed methods of modeling multiple aspects of human communication dynamics, in the context of applications in healthcare (depression, PTSD, suicide, autism), education (learning analytics), business (negotiation, interpersonal skills) and social multimedia (opinion mining, social in uence).

The next speaker, Dr. Daniel McDu from Microsoft AI, focused on Building Intelligent and Visceral Machines. The talk covered methods for physiological and behavioral measurement via ubiquitous hardware and then detailed the state-of-the-art approaches for synthesizing behavioral signals. The speaker led with examples of new human-computer interfaces and autonomous systems that leverage behavioral and physiological models, including a ect-aware natural language conversation systems, cross-domain learning systems, and vehicles with intrinsic emotional drives. This talk also included a discussion on ethics in the context of a ect-aware machines.

Dr. Natasha Jaques, from Google Brain, presented reinforcement learningbased methods leveraged to generate a ective dialogues. The methodology presented here was a smart application of applying RL by codifying soft concepts such as feelings and a ect. The method leveraged transfer learning to ne-tune a pre-trained dialog model with human feedback using reinforcement learning, and shows how learning from cues like a user's sentiment is more e ective than relying on manual labels. These techniques were applied to applications that learn novel conversational rewards, including reducing the toxicity of language generated by the model.

The next session focused on the marketing science perspective of interactions. Prof. Tom Novak and Prof. Donna Ho man from George Washington University presented a machine learning-based approach in the context of IoT and real-world data. They presented a computational approach that enabled operationalization and visualization of an assemblage theory interpretation of the emergence of automation practices in the Internet of Things. Their approach created a representation of the possibility space of automation assemblages that revealed the boundaries of territorialized automation practices and used this representation as a basis for qualitative analysis, theory development, and estimates of future growth. Extending these methodologies towards a ective interactions is an exciting research space. Their keynote was followed by an invited talk from Alain Lemaire from the University of Columbia, who presented his work with Prof. Netzer on linguistic matching of products and consumers. Their empirical analysis suggests that preferences for products can be inferred from the similarity between prospective customers' linguistic style, as well as the language used by other customers to describe a product.

Prof. Robert Kraut, from CMU, presented the nal keynote. Dr. Kraut contributed a social psychology perspective to the discussion of interactive a ect, covering aspects of social agency and support in support groups. An automated analysis that studied the interactions and dependency patterns in support groups was discussed. An insightful study around understanding how the language in these sites in uences how long people stay, the support they receive, and their satisfaction with it. These ndings form the basis of methods of interventions to better match support providers with support recipients in both online and in-person support groups environments. 3.2 The workshop included 4 full paper presentations and 4 posters. Lin et al. [ 40 ] presented their work on context-dependent models for facial expression prediction. Their method aimed to predict expressiveness from visual signals. The models beat baselines and perform at par with human annotations in terms of correlation with the ground truth.

Schoene et al. [ 75 ] presented their work on a bidirectional LSTM-based model for Fine-grained emotion classi cation in Tweets. Their approach showed that a dilated Bi-LSTM with attention achieved state-of-the-art performances beating automated baselines for multiple datasets. The method also outperforms human benchmarks for the emotion classi cation task.

Fong and Kumar [ 26 ] presented an interesting paper based on a hierarchical approach for emotion classi cation. They present baseline models for both coarse and ne-grained emotion classi cation. This paper presented a novel 24-way classi cation scheme for emotion classes. The results showed that the proposed models outperform other baselines across various classes.

Chen et al. [ 13 ] presented a fusion-based approach for multi-feature, multimodal sentiment analysis. Their approach combined audio and text features for sentiment classi cation. They report state-of-the-art results on the IEMOCAP database for multimodal emotion recognition. 3.3

Posters

Four posters were accepted to the workshop. Two of these posters leveraged audio / music data for a ective analysis [ 25,12 ], Xu et al. [ 93 ] modeled customer needs using sentiment based model, and Bara et al. [ 2 ] presented an approach for stress detection using multi-modal data. 3.4

CL-A

Shared Task

Another highlight of the workshop was the 2nd Computational Linguistics Affect Understanding (CL-A ) Shared Task on modeling interactive a ective responses. A new dataset, titled the O MyChest dataset, was released alongwith two complementary challenges to model disclosure and supportive behavior in social media discussions. Six teams participated in the task. An overview of the approaches and the results is provided as a part of this proceedings [ 32 ]. The system approaches were presented as a part of the poster session and are included in this proceedings volume [ 74,31,83,58,1,92 ]. 4

Related Workshops

Many workshops and conferences are now exploring the problems around a ective computing. This suggests the increasing importance of the research problem and the timeliness of this workshop for the AI community. The following workshops focused mainly on text analysis, sentiment, and subjectivity of the text content: { SENTIRE series: The workshop on Sentiment Elicitation from Natural Text for Information Retrieval and Extraction has been a continuing series for the past few years at ICDM 5. The organizers of this workshop series are part of the program committee for the proposed workshop. { WASSA: The workshop on Computational Approaches to Subjectivity, Sentiment & Social Media Analysis is a workshop series that concentrates on sentiment analysis in text and looks at various aspect{based and subjectivity analysis of text in that context. The workshop has been a popular workshop at top NLP conferences such as EMNLP, ACL, and NAACL in recent years 6. The organizers of this workshop series as well are a part of the program committee of this proposed workshop.

The following workshops focused on multimodal sensory data in their analysis. Text and language analysis is, however, not the focus of these workshops. This makes the AAAI Workshop on A ective Content Analysis rather unique in its pitch to bring the two communities together.

{ The rst workshop on A ective Computing (IJCAI 2017) concentrates on measuring human a ect based on sensors and wearable devices. { 1st Workshop on Tools and Algorithms for Mental Health and Wellbeing,

Pain, and Distress (MHWPD) { Multimodal Emotion Recognition Challenge (MEC 2017) @ 2018 Asian Conference on A ective Computing and Intelligent Interaction (AACII) Other current relevant events include ACII7, HUMANAIZE8, and NLP+CSS9. 5

Outlook

This workshop received a promising number of submissions and generated a lot of interest from scholars and industry. The response to the Shared Task was also successful at identifying a community of researchers and a variety of resources for a ect analysis in text. The program comprising interdisciplinary keynotes, original research presentations, a poster session, and a Shared Task has proven to be a successful and agile format. We will continue this multi{disciplinary workshop in an attempt to establish the space of computational approaches for a ective content analysis.

Acknowledgments

We want to thank Adobe Research for their generous funding, which made this workshop possible. We thank our program committee members who did an ex5 http://sentic.net/sentire/ 6 http://optima.jrc.it/wassa2017/ 7 http://acii2017.org/ 8 http://st.sigchi.org/publications/toc/humanize-2017.html 9 https://sites.google.com/site/nlpandcss/nlp-css-at-acl-2017 cellent job of reviewing the submissions. All PC members are documented on the A Con-19 website10. 10 https://sites.google.com/view/a con2019/committees main conference and the shared task, and Volume 2: Proceedings of the Sixth International Workshop on Semantic Evaluation, Association for Computational Linguistics (2012) 246{255 52. Mohammad, S.M., Turney, P.D.: Emotions evoked by common words and phrases: Using mechanical turk to create an emotion lexicon. In: Proceedings of the NAACL HLT 2010 workshop on computational approaches to analysis and generation of emotion in text, Association for Computational Linguistics (2010) 26{34 53. Mohammad, S.M., Turney, P.D.: Crowdsourcing a word-emotion association lexicon. 29(3) (2013) 436{465 54. Mohammad, S.M., Turney, P.D.: Nrc emotion lexicon. Technical report, NRC

Technical Report (2013) 55. Muhammad, A., Wiratunga, N., Lothian, R., Glassey, R.: Domain-based lexicon enhancement for sentiment analysis. In: SMA@ BCS-SGAI. (2013) 7{18 56. Nester, D., Haduong, N., Vaidyanathan, P., Prud'Hommeaux, E., Bailey, R., Alm, C.: Multimodal Alignment for A ective Content. In: Proceedings of the AAAI-18 Workshop on A ective Content Analysis, New Orleans, USA, AAAI (2018) 57. Nielsen, F.A.: A new anew: Evaluation of a word list for sentiment analysis in microblogs. arXiv preprint arXiv:1103.2903 (2011) 58. Pant, K., Dadu, T., Mamidi, R.: Bert-based ensembles for modeling disclosure and support in conversational social media text. In: Proceedings of the 3rd Workshop on A ective Content Analysis @ AAAI (A Con2020), New York, New York (February 2020) 59. Pavlick, E., Tetreault, J.: An empirical analysis of formality in online communication. Transactions of the Association for Computational Linguistics 4 (2016) 61{74 60. Pennebaker, J.W., Francis, M.E., Booth, R.J.: Linguistic inquiry and word count:

Liwc 2001. Mahway: Lawrence Erlbaum Associates 71 (2001) 2001 61. Peterson, K., Hohensee, M., Xia, F.: Email formality in the workplace: A case study on the enron corpus. In: Proceedings of the Workshop on Languages in Social Media, Association for Computational Linguistics (2011) 86{95 62. Picard, R.W., Picard, R.: A ective computing. Volume 252. MIT press Cambridge (1997) 63. Pittman, M., Reich, B.: Social media and loneliness: Why an instagram picture may be worth more than a thousand twitter words. Computers in Human Behavior 62 (2016) 155{167 64. Plutchik, R.: A general psychoevolutionary theory of emotion. Theories of Emotion 1 (1980) 3{31 65. Preotiuc-Pietro, D., Xu, W., Ungar, L.H.: Discovering user attribute stylistic differences via paraphrasing. In: Proceedings of the Thirtieth AAAI Conference on Arti cial Intelligence, February 12-17, 2016, Phoenix, Arizona, USA. (2016) 3030{ 3037 66. Qiu, G., Liu, B., Bu, J., Chen, C.: Opinion word expansion and target extraction through double propagation. Computational linguistics 37(1) (2011) 9{27 67. Qiu, J., Chen, X., Hu, K.: A Novel Machine Learning-based Sentiment Analysis Method for Chinese Social Media Considering Chinese Slang Lexicon and Emoticons. In: Proceedings of the AAAI-19 Workshop on A ective Content Analysis, Honolulu, USA, AAAI (2019) 68. Rahimtoroghi, E., Wu, J., Wang, R., Anand, P., Walker, M.A.: Modelling protagonist goals and desires in rst-person narrative. arXiv preprint arXiv:1708.09040 (2017)

References

1. Akiti , C. , Rajtmajer , S. , Squicciarini , A. : Contextual representation of selfdisclosure and supportiveness in short text . In: Proceedings of the 3rd Workshop on A ective Content Analysis @ AAAI (A Con2020) , New York, New York ( February 2020 )

2. Bara , C.P. , Papakostas , M. , Mihalcea , R.: A Deep Learning Approach Towards Multimodal Stress Detection . In: Proceedings of the AAAI-20 Workshop on A ective Content Analysis , New York, USA, AAAI ( 2020 )

3. Berry , M.W. , Browne , M. , Signer , B. : Topic annotated enron email data set . Philadelphia: Linguistic Data Consortium ( 2001 )

4. Bestgen , Y. : Building a ective lexicons from speci c corpora for automatic sentiment analysis . In: LREC . ( 2008 )

5. Bestgen , Y. , Vincze , N.: Checking and bootstrapping lexical norms by means of word similarity indexes . Behavior Research Methods 44 ( 4 ) ( 2012 ) 998 { 1006

6. Bird , S. : Nltk: The natural language toolkit . In: Proceedings of the COLING/ACL on Interactive presentation sessions, Association for Computational Linguistics ( 2006 ) 69 { 72

7. Blei , D.M. , Ng , A.Y. , Jordan , M.I. : Latent dirichlet allocation . Journal of Machine Learning Research 3(Jan) ( 2003 ) 993 { 1022

8. Blitzer , J. , Dredze , M. , Pereira , F. , et al.: Biographies, bollywood, boom-boxes and blenders: Domain adaptation for sentiment classi cation . In: ACL . Volume 7 . ( 2007 ) 440 { 447

9. Bradley , M.M. , Lang , P.J.: A ective norms for english words (anew): Instruction manual and a ective ratings . Technical report, Technical report C-1 , the center for research in psychophysiology, University of Florida ( 1999 )

10. Brett , D. , Pinna , A. : The distribution of a ective words in a corpus of newspaper articles . Procedia-Social and Behavioral Sciences 95 ( 2013 ) 621 { 629

11. Cambria , E. , Fu , J. , Bisio , F. , Poria , S.: A ectivespace 2: Enabling a ective intuition for concept-level sentiment analysis . In: AAAI . ( 2015 ) 508 { 514

12. Chaki , S. , Doshi , P. , Patnaik , P. , Bhattacharya , S. : Attentive RNNs for Continuoustime Emotion Prediction in Music Clips . In: Proceedings of the AAAI-20 Workshop on A ective Content Analysis , New York, USA, AAAI ( 2020 )

13. Chen , F. , Luo , Z. , Xu , Y. , Ke , D. : Complementary Fusion of Multi-Features and Multi-Modalities in Sentiment Analysis . In: Proceedings of the AAAI-20 Workshop on A ective Content Analysis , New York, USA, AAAI ( 2020 )

14. Chhaya , N. , Jaidka , K. , Ungar , L.H. : The AAAI-18 Workshop on A ective Content Analysis . In: Proceedings of the AAAI-18 Workshop on A ective Content Analysis , New Orleans, USA, AAAI ( 2018 )

15. Chikersal , P. , Poria , S. , Cambria , E. , Gelbukh , A. , Siong , C.E. : Modelling public sentiment in twitter: using linguistic patterns to enhance supervised learning . In: International Conference on Intelligent Text Processing and Computational Linguistics , Springer ( 2015 ) 49 { 65

16. Cohen , W.W. : Enron email dataset . ( 2009 )

17. Colombetti , G.: From a ect programs to dynamical discrete emotions . Philosophical Psychology 22 ( 4 ) ( 2009 ) 407 { 425

18. Danescu-Niculescu-Mizil , C. , Sudhof , M. , Jurafsky , D. , Leskovec , J. , Potts , C. : A computational approach to politeness with application to social factors . arXiv preprint arXiv:1306.6078 ( 2013 )

19. Derakhshan , A. , Mikaeili , M. , Gedeon , T. : Discriminating Between Truthfulness and Deception Using Infrared Thermal Imaging and Peripheral Physiology . ( 2018 )

20. Ding , H. , Jiang , T. , Rilo , E.: Why is an Event A ective? Classifying A ective Events based on Human Needs . In: Proceedings of the AAAI-18 Workshop on A ective Content Analysis , New Orleans, USA, AAAI ( 2018 )

21. Dragut , E.C. , Yu , C. , Sistla , P. , Meng , W. : Construction of a sentimental word dictionary . In: Proceedings of the 19th ACM International Conference on Information and Knowledge Management. CIKM '10 , New York, NY, USA, ACM ( 2010 ) 1761 { 1764

22. Dumpala , S.H. , Chakraborty , R. , Kopparapu , S.K. : Knowledge driven feed-forward neural network for audio a ective content analysis . In: Proceedings of the AAAI-18 Workshop on A ective Content Analysis , New Orleans, USA, AAAI ( 2018 )

23. Ekman , P.: An argument for basic emotions . Cognition & Emotion 6 ( 3 -4) ( 1992 ) 169 { 200

24. Esuli , A. , Sebastiani , F. : Sentiwordnet: A publicly available lexical resource for opinion mining . In: Proceedings of LREC. Volume 6 ., Citeseer ( 2006 ) 417 { 422

25. Feng , K. , Chaspari , T. : A Siamese Neural Network with Modi ed Distance Loss For Transfer Learning in Speech Emotion Recognition . In: Proceedings of the AAAI-20 Workshop on A ective Content Analysis , New York, USA, AAAI ( 2020 )

26. Fong , H. , Kumar , V.: A Hierarchical Approach to Machine Emotional Intelligence . In: Proceedings of the AAAI-20 Workshop on A ective Content Analysis , New York, USA, AAAI ( 2020 )

27. Hatzivassiloglou , V. , McKeown , K.R. : Predicting the semantic orientation of adjectives . In: Proceedings of the eighth conference on European chapter of the Association for Computational Linguistics, Association for Computational Linguistics ( 1997 ) 174 { 181

28. He , S. , Ballard , D. , Gildea , D. : Building and tagging with an a ect lexicon . ( 2004 )

29. Hu , C. , Walker , M.A. , Ne , M. , Tree , J.E.F. : Storytelling agents with personality and adaptivity . In: International Conference on Intelligent Virtual Agents , Springer ( 2015 ) 181 { 193

30. Huang , C. r.: Ontology and the Lexicon: A Natural Language Processing Perspective . Cambridge University Press ( 2010 )

31. Hyun , J. , Bae , B.C. , Cheong , Y.G. : [cl-a shared task] multi-label text classi cation using an emotion embedding model . In: Proceedings of the 3rd Workshop on A ective Content Analysis @ AAAI (A Con2020) , New York, New York ( February 2020 )

32. Jaidka , K. , Singh , I. , Lu , J. , Chhaya , N. , Ungar , L. : A report of the CL-A O MyChest Shared Task: Modeling Supportiveness and Disclosure . In: Proceedings of the AAAI-20 Workshop on A ective Content Analysis , New York, USA, AAAI ( 2020 )

33. Jones , E. , Oliphant , T. , Peterson , P. , et al.: SciPy: Open source scienti c tools for Python ( 2001 {) [Online; accessed 2017- 01 -17].

34. Joshi , T. , Sivaprasad , S. , Pedanekar , N.: Partners in Crime: Utilizing ArousalValence Relationship for Continuous Prediction of Valence in Movies . In: Proceedings of the AAAI-19 Workshop on A ective Content Analysis , Honolulu, USA, AAAI ( 2019 )

35. Kanayama , H. , Nasukawa , T. : Fully automatic lexicon expansion for domainoriented sentiment analysis . In: Proceedings of the 2006 conference on empirical methods in natural language processing , Association for Computational Linguistics ( 2006 ) 355 { 363

36. Kanayama , H. , Nasukawa , T. : Fully automatic lexicon expansion for domainoriented sentiment analysis . In: Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing. EMNLP '06 , Stroudsburg , PA, USA, Association for Computational Linguistics ( 2006 ) 355 { 363

37. Kowalczyk , D. , Larsen , J.: Scalable Privacy-Compliant Virality Prediction on Twitter. In: Proceedings of the AAAI-19 Workshop on A ective Content Analysis , Honolulu, USA, AAAI ( 2019 )

38. Le , H. , Cerisara , C. , Denis , A. : Do convolutional networks need to be deep for Text Classi cation . In: Proceedings of the AAAI-18 Workshop on A ective Content Analysis , New Orleans, USA, AAAI ( 2018 )

39. Li , D. , Rzepka , R. , Ptaszynski , M. , Araki , K. : Audio Sentiment Analysis by Heterogeneous Signal Features Learned from Utterance-Based Parallel Neural Network . In: Proceedings of the AAAI-19 Workshop on A ective Content Analysis , Honolulu, USA, AAAI ( 2019 )

40. Lin , V. , Girard , J. , Morency , L.P. : Context-Dependent Models for Predicting and Characterizing Facial Expressiveness . In: Proceedings of the AAAI-20 Workshop on A ective Content Analysis , New York, USA, AAAI ( 2020 )

41. Litvinova , T. , Litvinova , O. , Seredin , P. , Zagorovskaya , O. : RusNeuroPsych: Corpus for Study Relations Between Author Demo-graphic, Personality Traits, Lateral Preferences and A ect in Text . ( 2018 )

42. Liu , T. , Kappas , A. : Predicting Engagement Breakdown in HRI Using Thin-slices of Facial Expressions . In: Proceedings of the AAAI-18 Workshop on A ective Content Analysis , New Orleans, USA, AAAI ( 2018 )

43. Lu , Y. , Castellanos , M. , Dayal , U. , Zhai , C. : Automatic construction of a contextaware sentiment lexicon: An optimization approach . In: Proceedings of the 20th International Conference on World wide web, ACM ( 2011 ) 347 { 356

44. Luo , Z. , Xu , H. , Chen , F. : Audio Sentiment Analysis by Heterogeneous Signal Features Learned from Utterance-Based Parallel Neural Network . In: Proceedings of the AAAI-19 Workshop on A ective Content Analysis , Honolulu, USA, AAAI ( 2019 )

45. Maheshwari , T. , Reganti , A.N. , Kumar , U. , Chakraborty , T. , Das , A. : Semantic interpretation of social network communities . In: AAAI . ( 2017 ) 4967 { 4968

46. Mairesse , F. , Walker , M.A. : Trainable generation of big- ve personality styles through data-driven parameter estimation . In: ACL . ( 2008 ) 165 { 173

47. Manning , C.D. , Surdeanu , M. , Bauer , J. , Finkel , J.R. , Bethard , S. , McClosky , D. : The stanford corenlp natural language processing toolkit . In: ACL (System Demonstrations) . ( 2014 ) 55 { 60

48. Mehrabian , A. : Basic dimensions for a general psychological theory implications for personality, social, environmental, and developmental studies . ( 1980 )

49. Miller , G.A. : Wordnet: a lexical database for english . Communications of the ACM 38 ( 11 ) ( 1995 ) 39 { 41

50. Mitchell, S. , OSullivan, M. , Dunning , I. : Pulp: A linear programming toolkit for python . The University of Auckland, Auckland, New Zealand, http://www. optimization-online. org/DB FILE/ 2011 /09/3178. pdf ( 2011 )

51. Mohammad , S.M. : # emotional tweets . In: Proceedings of the First Joint Conference on Lexical and Computational Semantics-Volume 1 : Proceedings of the

69. Ranganath , R. , Jurafsky , D. , McFarland , D.A. : Detecting friendly, irtatious, awkward, and assertive speech in speed-dates . Computer Speech & Language 27 ( 1 ) ( 2013 ) 89 { 115

70. Reed , L. , Wu , J. , Oraby , S. , Anand , P. , Walker , M. : Learning lexico-functional patterns for rst-person a ect . arXiv preprint arXiv:1708.09789 ( 2017 )

71. Ribeiro , F.N. , Araujo , M. , Goncalves , P. , Goncalves , M.A. , Benevenuto , F. : Sentibench-a benchmark comparison of state-of-the-practice sentiment analysis methods . EPJ Data Science 5 ( 1 ) ( 2016 ) 1 { 29

72. Robertson , S. : Understanding inverse document frequency: On theoretical arguments for idf . Journal of Documentation 60 ( 5 ) ( 2004 ) 503 { 520

73. Sakaguchi , K. , Duh , K. , Post , M. , Van Durme , B. : Robsut wrod reocginiton via semi-character recurrent neural network . In: AAAI . ( 2017 ) 3281 { 3287

74. Saxon , M. , Bhandari , S. , Ruskin , L. , Honda , G.: Word pair convolutional model for happy moment classi cation . In: Proceedings of the 2nd Workshop on A ective Content Analysis @ AAAI (A Con2019) , Honolulu, Hawaii ( January 2019 )

75. Schoene , A.M. , Turner , A. , Dethlefs , N.: Bidirectional Dilated LSTM with Attention for Fine-grained EmotionClassi cation in Tweets . In: Proceedings of the AAAI-20 Workshop on A ective Content Analysis , New York, USA, AAAI ( 2020 )

76. Schwartz , H.A. , Eichstaedt , J.C. , Kern , M.L. , Dziurzynski , L. , Ramones , S.M. , Agrawal , M. , Shah , A. , Kosinski , M. , Stillwell , D. , Seligman , M.E. , et al.: Personality, gender, and age in the language of social media: The open-vocabulary approach . PloS One 8 ( 9 ) ( 2013 ) e73791

77. Seligman , M.E. : Flourish: a visionary new understanding of happiness and wellbeing . Policy 27 ( 3 ) ( 2011 ) 60 { 1

78. Stackover ow: Finding the best trade-o point on a curve ( 2015 )

79. Stone , P.J. , Dunphy , D.C. , Smith , M.S.: The general inquirer: A computer approach to content analysis . ( 1966 )

80. Strapparava , C. , Mihalcea , R.: Semeval-2007 task 14: A ective text . In: Proceedings of the 4th International Workshop on Semantic Evaluations, Association for Computational Linguistics ( 2007 ) 70 { 74

81. Strapparava , C. , Valitutti , A. , et al.: Wordnet a ect: an a ective extension of wordnet . In: LREC . Volume 4 . ( 2004 ) 1083 { 1086

82. Sun , B. , Yang , L. , Dong , P. , Zhang, W. , Dong , J. , Young , C. : Super Characters: A Conversion from Sentiment Classi cation to Image Classi cation . In: Proceedings of the AAAI-19 Workshop on A ective Content Analysis , Honolulu, USA, AAAI ( 2019 )

83. Sun , B. , Yang , L. , Sha , H. , Lin , M.: Multi-modal sentiment analysis using super characters method on low-power cnn accelerator device . In: Proceedings of the 3rd Workshop on A ective Content Analysis @ AAAI (A Con2020) , New York, New York ( February 2020 )

84. Taboada , M. , Brooke , J. , To loski, M., Voll , K. , Stede , M. : Lexicon-based methods for sentiment analysis . Computational Linguistics 37 ( 2 ) ( 2011 ) 267 { 307

85. Thelwall , M. , Buckley , K. , Paltoglou , G. , Skowron , M. , Garcia , D. , Gobron , S. , Ahn , J. , Kappas , A. , Kuster, D. , Holyst , J.A. : Damping sentiment analysis in online communication: discussions, monologs and dialogs . In: International Conference on Intelligent Text Processing and Computational Linguistics , Springer ( 2013 ) 1 { 12

86. Tiam-Lee , J. , Sumi , K. : Emotional Experience of Students Interacting with a System for Learning Programming . In: Proceedings of the AAAI-19 Workshop on A ective Content Analysis , Honolulu, USA, AAAI ( 2019 )

87. Tighe , E.P. , Ureta , J.C. , Pollo , B.A.L. , Cheng, C.K., de Dios Bulos , R.: Personality trait classi cation of essays with the application of feature reduction . In: SAAIP@ IJCAI . ( 2016 ) 22 { 28

88. Turney , P.D. , Littman , M.L. : Measuring praise and criticism: Inference of semantic orientation from association . ACM Transactions on Information Systems (TOIS) 21(4) ( 2003 ) 315 { 346

89. Warriner , A.B. , Kuperman , V. , Brysbaert , M. : Norms of valence, arousal, and dominance for 13,915 english lemmas . Behavior research methods 45(4) ( 2013 ) 1191 { 1207

90. Wiebe , J. , Wilson, T. , Bell , M. : Identifying collocations for recognizing opinions . In: Proceedings of the ACL-01 Workshop on Collocation: Computational Extraction , Analysis, and Exploitation. ( 2001 ) 24 { 31

91. Wu , J. , Walker , M. , Anand , P. , Whittaker , S. : Linguistic re exes of well-being and happiness in echo . In: Proceedings of the 8th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis. ( 2017 ) 81 { 91

92. Xin , W. , Inkpen , D. : [cl-a shared task] detecting disclosure and support via deep multi-task learning . In: Proceedings of the 3rd Workshop on A ective Content Analysis @ AAAI (A Con2020) , New York, New York ( February 2020 )

93. Xu , J. , Tang , B. , Liu , X. , Xiong , F. : Automatically Identifying Customer Needs with Attribute Sentiment Model . In: Proceedings of the AAAI-20 Workshop on A ective Content Analysis , New York, USA, AAAI ( 2020 )

94. Zahiri , S. , Choi , J.: Emotion Detection on TV Show Transcripts with Sequencebased Convolutional Neural Networks . In: Proceedings of the AAAI-18 Workshop on A ective Content Analysis , New Orleans, USA, AAAI ( 2018 )

95. Zhao , S. , Yao , H. , Jiang , X. : Predicting continuous probability distribution of image emotions in valence-arousal space . In: Proceedings of the 23rd ACM International Conference on Multimedia. MM '15 , New York, NY, USA, ACM ( 2015 ) 879 { 882