由效價值（valence）和喚醒度（arousal）構成的二維空間，是一種維度型情緒定義方法。相較於類別型方法，維度型方法更加適合做定量的研究分析。在進行維度型方法情緒研究的過程中，產生了許多標註者對文字的效價值和喚醒度的標註。研究者發現，除了文字本身，還有一些其他會對標註者的標註產生影響的因素，例如標註者的年紀差異，標註者的文化差異，標註者的個性差異。本研究用五大個人模型，來探索由於標註者個性差異而對文字進行效價值和喚醒度進行標註時的差異，結果顯示標註者個性的不同維度的差異會對不同情緒的文字標註產生不同的影響。同時，本研究將個性作為一個新的考慮因素加入到效價值和喚醒度的預測模型中，提升了該預測模型的表現。

The two-dimensional space, which consists of valence and arousal, is a dimensional method for emotion analysis. Compared to traditional determinative emotion classification method, the dimensional method is more suitable for quantitative analysis. A lot of annotations were used in valence and arousal space related researches. From these annotations, researchers noticed that besides texts, some features of annotators may also have some influences on annotation, like age, culture and personality. This research intends to use Big Five personality traits discover the personality-related differences of valence and arousal annotations, and the results show that different factors dimensions of the Big Five have different influences on VA annotations in different emotions. In the meantime, this research also intends to verify whether considering personality as a new feature will improve the performance of VA prediction or not.

1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1
2 Related Work. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3
2.1Emotion Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3
2.2Personality Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4
2.3Annotation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6
3 Methodology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7
3.1Valence and Arousal . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7
3.2The Big Five and the Big Five Inventory . . . . . . . . . . . . . . . . . . .9
3.3The T-test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.4EmoPredictor with Personality . . . . . . . . . . . . . . . . . . . . . 12
4 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
4.1Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.1.1Texts Selecting . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.1.2BFI scoring and VA annotation . . . . . . . . . . . . . . . . . . . . 15
4.2Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2.1Neuroticism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2.2Extraversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.2.3Openness to Experience . . . . . . . . . . . . . . . . . . . . . . . 27
4.2.4Agreeableness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.2.5Conscientiousness . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.3Evaluation for EmoPredictor with Personality . . . . . . . . . . . . . 39
4.3.1Evaluation for Modeling . . . . . . . . . . . . . . . . . . . . . . . 39
4.3.2Evaluation for Prediction . . . . . . . . . . . . . . . . . . . . . . . 39
5 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .41
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43

[1] Cem Akkaya, Alexander Conrad, Janyce Wiebe, and Rada Mihalcea. Amazon me-chanical turk for subjectivity word sense disambiguation. InProceedings of theNAACL HLT 2010 workshop on creating speech and language data with Amazon’sMechanical Turk, pages 195–203. Association for Computational Linguistics, 2010.
[2] Carlos Argueta and Yi-Shin Chen. Multi-lingual sentiment analysis of social databased on emotion-bearing patterns.SocialNLP 2014, page 38, 2014.
[3] Richard W Backs, Sergio P da Silva, and Kyunghee Han. A comparison of youngerand older adults’ self-assessment manikin ratings of affective pictures.Experimentalaging research, 31(4):421–440, 2005.
[4] Eugene Yuta Bann and Joanna J Bryson. Measuring cultural relativity of emo-tional valence and arousal using semantic clustering and twitter.arXiv preprintarXiv:1304.7507, 2013.[5] Veronica Benet-Martinez and Oliver P John. Los cinco grandes across cultures andethnic groups: Multitrait-multimethod analyses of the big five in spanish and english.Journal of personality and social psychology, 75(3):729, 1998.43
[6] Alberto Betella and Paul FMJ Verschure.The affective slider: a digital self-assessment scale for the measurement of human emotions.PloS one, 11(2):e0148037,2016.
[7] Margaret Bradley and Peter J Lang.The International affective digitized sounds(IADS)[: stimuli, instruction manual and affective ratings. NIMH Center for theStudy of Emotion and Attention, 1999.
[8] Margaret M Bradley and Peter J Lang. Measuring emotion: the self-assessmentmanikin and the semantic differential.Journal of behavior therapy and experimentalpsychiatry, 25(1):49–59, 1994.
[9] Margaret M Bradley and Peter J Lang. Affective norms for english words (anew):Instruction manual and affective ratings. Technical report, Citeseer, 1999.
[10] Michael Buhrmester, Tracy Kwang, and Samuel D Gosling. Amazon’s mechanicalturk: A new source of inexpensive, yet high-quality, data?Perspectives on psycho-logical science, 6(1):3–5, 2011.
[11] Arnold H Buss. Personality as traits.American Psychologist, 44(11):1378, 1989.
[12] Chris Callison-Burch. Fast, cheap, and creative: evaluating translation quality us-ing amazon’s mechanical turk. InProceedings of the 2009 Conference on EmpiricalMethods in Natural Language Processing: Volume 1-Volume 1, pages 286–295. As-sociation for Computational Linguistics, 2009
[13] Philip J Corr and Gerald Matthews.The Cambridge handbook of personality psychol-ogy. Cambridge University Press New York, 2009.
[14] Paul T Costa and Robert R MacCrae.Revised NEO personality inventory (NEO PI-R) and NEO five-factor inventory (NEO-FFI): Professional manual. PsychologicalAssessment Resources, Incorporated, 1992.
[15] Paul T Costa and Robert R McCrae. Normal personality assessment in clinical prac-tice: The neo personality inventory.Psychological assessment, 4(1):5, 1992.
[16] PT Costa and RR McCrae. Neo five-factor inventory (neo-ffi).Odessa, FL: Psycho-logical Assessment Resources, 1989.
[17] Tuomas Eerola and Jonna K Vuoskoski. A comparison of the discrete and dimensionalmodels of emotion in music.Psychology of Music, 39(1):18–49, 2011.
[18] Paul Ekman, E Richard Sorenson, Wallace V Friesen, et al. Pan-cultural elements infacial displays of emotion.Science, 164(3875):86–88, 1969.
[19] Nico H Frijda. The emotions cambridge university press cambridge.UK GoogleScholar, 1986.
[20] Lewis R Goldberg. The development of markers for the big-five factor structure.Psychological assessment, 4(1):26, 1992.
[21] Lewis R Goldberg, John A Johnson, Herbert W Eber, Robert Hogan, Michael C Ash-ton, C Robert Cloninger, and Harrison G Gough. The international personality itempool and the future of public-domain personality measures.Journal of Research inpersonality, 40(1):84–96, 2006.
[22] Daniel Gr ̈uhn and Susanne Scheibe. Age-related differences in valence and arousal ratings of pictures from the international affective picture system (iaps): Do ratingsbecome more extreme with age?Behavior Research Methods, 40(2):512–521, 2008.
[23] Oliver P John, Laura P Naumann, and Christopher J Soto. Paradigm shift to theintegrative big five trait taxonomy.Handbook of personality: Theory and research,3:114–158, 2008.
[24] Oliver P John and Sanjay Srivastava. The big five trait taxonomy: History, measure-ment, and theoretical perspectives.Handbook of personality: Theory and research,2(1999):102–138, 1999.
[25] Suin Kim, JinYeong Bak, and Alice Haeyun Oh. Do you feel what i feel? socialaspects of emotions in twitter conversations. InICWSM, 2012.
[26] Yoon Kim. Convolutional neural networks for sentence classification.arXiv preprintarXiv:1408.5882, 2014.
[27] Efthymios Kouloumpis, Theresa Wilson, and Johanna D Moore. Twitter sentimentanalysis: The good the bad and the omg!Icwsm, 11(538-541):164, 2011.
[28] Miiamaaria V Kujala, Sanni Somppi, Markus Jokela, Outi Vainio, and Lauri Parkko-nen. Human empathy, personality and experience affect the emotion ratings of dogand human facial expressions.PloS one, 12(1):e0170730, 2017
[29] Peter Kuppens, Francis Tuerlinckx, Michelle Yik, Peter Koval, Joachim Coosemans,Kevin J Zeng, and James A Russell. The relation between valence and arousal insubjective experience varies with personality and culture.Journal of personality,85(4):530–542, 2017.
[30] Peter J Lang, Margaret M Bradley, and Bruce N Cuthbert. International affectivepicture system (iaps): Technical manual and affective ratings.NIMH Center for theStudy of Emotion and Attention, pages 39–58, 1997.
[31] Gerald Matthews, Ian J Deary, and Martha C Whiteman.Personality traits. Cam-bridge University Press, 2003.
[32] Alexander Pak and Patrick Paroubek. Twitter as a corpus for sentiment analysis andopinion mining. InLREc, volume 10, 2010.
[33] Bo Pang, Lillian Lee, et al. Opinion mining and sentiment analysis.Foundations andTrendsR©in Information Retrieval, 2(1–2):1–135, 2008.
[34] Robert Plutchik. The nature of emotions human emotions have deep evolutionaryroots, a fact that may explain their complexity and provide tools for clinical practice.American scientist, 89(4):344–350, 2001.
[35] Daniel Preot ̧iuc-Pietro, H Andrew Schwartz, Gregory Park, Johannes Eichstaedt,Margaret Kern, Lyle Ungar, and Elisabeth Shulman. Modelling valence and arousalin facebook posts. InProceedings of the 7th Workshop on Computational Approachesto Subjectivity, Sentiment and Social Media Analysis, pages 9–15, 2016.
[36] David L Robinson. Brain function, emotional experience and personality.Nether-lands Journal of Psychology, 64(4):152–168, 2008.
[37] James A. Russel. A circumplex model of affect.Journal of Personality and SocialPsychology, 39(6):1161–1178, 1980.
[38] Graeme D Ruxton. The unequal variance t-test is an underused alternative to student’st-test and the mann–whitney u test.Behavioral Ecology, 17(4):688–690, 2006.
[39] Klaus R Scherer. What are emotions? and how can they be measured?Social scienceinformation, 44(4):695–729, 2005.
[40] Jinping Shi and Yi-Shin Chen. Quantitative emotion prediction of texts in valenceand arousal space from microblog data. 2017.
[41] Svitlana Volkova, Theresa Wilson, and David Yarowsky. Exploring sentiment in so-cial media: Bootstrapping subjectivity clues from multilingual twitter streams. InACL (2), pages 505–510, 2013.
[42] Hao Wang, Dogan Can, Abe Kazemzadeh, Franc ̧ois Bar, and Shrikanth Narayanan. Asystem for real-time twitter sentiment analysis of 2012 us presidential election cycle.InProceedings of the ACL 2012 System Demonstrations, pages 115–120. Associationfor Computational Linguistics, 2012.