最近的研究表明，小組成員之間的個性複合效應與整體小組績效有關，整合小組級個性和聲音行為可以增強在小組互動過程預測任務績效的預測能力。在預測小組任務績效時，我們建議應該從組內和組間兩個角度來建模小組成員的個性對於聲音行為的影響。具體來說，我們提出了一種基於圖學的複合型神經網絡：圖像對話聲學神經網絡（G-IAN）架構，用於建模小組級別的個性組成與成員的聲學行為之間的組內效應；此外，它利用組間的個性組成相似性生成圖表，以學習預測任務績效的組間關係。我們的結果顯示，G-IAN在NTULP和GGID數據庫上均實現了良好的預測效能，分類準確率在想個資料庫上分別達到了72.2％和78.4％，這比僅對聲音行為進行建模的基線模型絕對要高14％。我們也在ELEA資料庫上驗證了G-IAN的效能，並達到了64.97 MSE（0.409皮爾遜相關度）的預測效能，優於目前相關研究的預測效能。此外，我們在二維的人格空間進行可視化分析，以進一步闡明群體之間的結構關係。

Recent studies have indicated that the composite personality effect between group members is related to the overall group performance, the integration of group-level personality and vocal behaviors can provide enhanced prediction power on task performance for small group interactions. In this work, we propose that the impact of member personality for task performance prediction in groups should be explicitly modeled form both intra and inter-group perspectives. Specifically, we propose the Graph Interlocutor Acoustic Network (G-IAN) architecture that models the intra-group effect between group-level personality composition and members’ vocal behavior; additionally, it utilizes the similarity of personality composition between groups as the graph to learns the inter-group relationship in predicting the task performance. Our result shows that G-IAN achieves the promising performance, 72.2% and 78.4% task performance classification accuracy in both NTULP and GGID datasets, which outperforms the baseline model that models vocal behavior only by 14%; it also presents the state-of-art performance at 64.97 MSE better than the current best methods in ELEA dataset. Moreover, we present visualization analysis on the 2D personality space to further illuminate the structural relationship between groups.

[1] S. Tubbs, “A systems approach to small group interaction,” Manag. Fac. Scholarsh., Jan. 2012, [Online]. Available: https://commons.emich.edu/mgmt_facsch/68.
[2] S. T. Bell, “Deep-level composition variables as predictors of team performance: A meta-analysis.,” J. Appl. Psychol., vol. 92, no. 3, pp. 595–615, 2007, doi: 10.1037/0021-9010.92.3.595.
[3] A. Kramer, D. P. Bhave, and T. D. Johnson, “Personality and group performance: The importance of personality composition and work tasks,” Personal. Individ. Differ., vol. 58, pp. 132–137, Feb. 2014, doi: 10.1016/j.paid.2013.10.019.
[4] R. L. Moreland, J. M. Levine, and M. L. Wingert, “Creating the ideal group: Composition effects at work,” in Understanding group behavior, Vol. 2: Small group processes and interpersonal relations, Hillsdale, NJ, US: Lawrence Erlbaum Associates, Inc, 1996, pp. 11–35.
[5] E. E. Ghiselli and T. M. Lodahl, “Patterns of managerial traits and group effectiveness,” J. Abnorm. Soc. Psychol., vol. 57, no. 1, pp. 61–66, 1958, doi: 10.1037/h0040588.
[6] B. Bradley, A. Klotz, B. Postlethwaite, and K. Brown, “Ready to Rumble: How Team Personality Composition and Task Conflict Interact to Improve Performance,” J. Appl. Psychol., vol. 98, Sep. 2012, doi: 10.1037/a0029845.
[7] W. Haythorn, “The influence of individual members on the characteristics of small groups,” J. Abnorm. Soc. Psychol., vol. 48, no. 2, pp. 276–284, 1953, doi: 10.1037/h0058211.
[8] B. Barry and G. L. Stewart, “Composition, process, and performance in self-managed groups: the role of personality,” J. Appl. Psychol., vol. 82, no. 1, pp. 62–78, Feb. 1997, doi: 10.1037/0021-9010.82.1.62.
[9] J. E. McGrath, Social psychology: a brief introduction. New York: Holt, Rinehart and Winston, 1964.
[10] G. Murray and C. Oertel, “Predicting Group Performance in Task-Based Interaction,” in Proceedings of the 20th ACM International Conference on Multimodal Interaction, Boulder, CO, USA, Oct. 2018, pp. 14–20, doi: 10.1145/3242969.3243027.
[11] U. Avci and O. Aran, “Predicting the Performance in Decision-Making Tasks: From Individual Cues to Group Interaction,” IEEE Trans. Multimed., vol. 18, no. 4, pp. 643–658, Apr. 2016, doi: 10.1109/TMM.2016.2521348.
[12] J. E. Driskell, R. Hogan, and E. Salas, “Personality and group performance,” in Group processes and intergroup relations, Thousand Oaks, CA, US: Sage Publications, Inc, 1987, pp. 91–112.
[13] Y.-S. Lin and C.-C. Lee, “Using Interlocutor-Modulated Attention BLSTM to Predict Personality Traits in Small Group Interaction,” in Proceedings of the 20th ACM International Conference on Multimodal Interaction, Boulder, CO, USA, Oct. 2018, pp. 163–169, doi: 10.1145/3242969.3243001.
[14] S.-C. Zhong, Y.-S. Lin, C.-M. Chang, Y.-C. Liu, and C.-C. Lee, “Predicting Group Performances Using a Personality Composite-Network Architecture During Collaborative Task,” in Interspeech 2019, Sep. 2019, pp. 1676–1680, doi: 10.21437/Interspeech.2019-2087.
[15] M. Kompan and M. Bieliková, Social Structure and Personality Enhanced Group Recommendation. .
[16] M. Bordons and M. A. Zulueta, “Comparison of research team activity in two biomedical fields,” Scientometrics, vol. 40, no. 3, pp. 423–436, Nov. 1997, doi: 10.1007/BF02459290.
[17] T. N. Kipf and M. Welling, “SEMI-SUPERVISED CLASSIFICATION WITH GRAPH CONVOLUTIONAL NETWORKS,” p. 14, 2017.
[18] B. C. Wheeler and B. E. Mennecke, “The School of Business Policy Task Manual: Working Paper #92-524c,” Indiana University Graduate School of Business, Working Paper, 1992. Accessed: Jun. 01, 2020. [Online]. Available: https://scholarworks.iu.edu/dspace/handle/2022/13427.
[19] L. R. Goldberg, “The development of markers for the Big-Five factor structure,” Psychol. Assess., vol. 4, no. 1, pp. 26–42, Mar. 1992, doi: 10.1037/1040-3590.4.1.26.
[20] R.-H. Li and Y.-C. Chen, “The Development of a Shortened Version of IPIP Big Five Personality Scale and the Testing of Its Measurement Invariance between Middle-Aged and Older People,” J. Educ. Res. Dev., vol. 12, no. 4, pp. 87–120, Dec. 2016, doi: 10.6925/SCJ.
[21] L. Zheng, L. Goldberg, Y. Zheng, Y. Zhao, Y. Tang, and L. Liu, “Reliability and Concurrent Validation of the IPIP Big-Five Factor Markers in China: Consistencies in Factor Structure between Internet-Obtained Heterosexual and Homosexual Samples,” Personal. Individ. Differ., vol. 45, pp. 649–654, Nov. 2008, doi: 10.1016/j.paid.2008.07.009.
[22] F. Eyben et al., “The Geneva Minimalistic Acoustic Parameter Set (GeMAPS) for Voice Research and Affective Computing,” IEEE Trans. Affect. Comput., vol. 7, no. 2, pp. 190–202, Apr. 2016, doi: 10.1109/TAFFC.2015.2457417.
[23] F. Eyben, M. Wöllmer, and B. Schuller, “Opensmile: the munich versatile and fast open-source audio feature extractor,” in Proceedings of the international conference on Multimedia - MM ’10, Firenze, Italy, 2010, p. 1459, doi: 10.1145/1873951.1874246.
[24] S. Hochreiter and J. Schmidhuber, “Long Short-Term Memory.”
[25] J. Chung, C. Gulcehre, K. Cho, and Y. Bengio, “Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling,” ArXiv14123555 Cs, Dec. 2014, Accessed: Jun. 01, 2020. [Online]. Available: http://arxiv.org/abs/1412.3555.
[26] K. Cho, B. van Merrienboer, D. Bahdanau, and Y. Bengio, “On the Properties of Neural Machine Translation: Encoder-Decoder Approaches,” ArXiv14091259 Cs Stat, Oct. 2014, Accessed: Jun. 01, 2020. [Online]. Available: http://arxiv.org/abs/1409.1259.
[27] J. Bruna, W. Zaremba, A. Szlam, and Y. LeCun, “Spectral Networks and Locally Connected Networks on Graphs,” ArXiv13126203 Cs, May 2014, Accessed: Jun. 23, 2020. [Online]. Available: http://arxiv.org/abs/1312.6203.
[28] M. Henaff, J. Bruna, and Y. LeCun, “Deep Convolutional Networks on Graph-Structured Data,” ArXiv150605163 Cs, Jun. 2015, Accessed: Jun. 23, 2020. [Online]. Available: http://arxiv.org/abs/1506.05163.
[29] D. K. Duvenaud et al., “Convolutional Networks on Graphs for Learning Molecular Fingerprints,” in Advances in Neural Information Processing Systems 28, C. Cortes, N. D. Lawrence, D. D. Lee, M. Sugiyama, and R. Garnett, Eds. Curran Associates, Inc., 2015, pp. 2224–2232.
[30] Y. Li, D. Tarlow, M. Brockschmidt, and R. Zemel, “Gated Graph Sequence Neural Networks,” ArXiv151105493 Cs Stat, Sep. 2017, Accessed: Jun. 23, 2020. [Online]. Available: http://arxiv.org/abs/1511.05493.
[31] M. Defferrard, X. Bresson, and P. Vandergheynst, “Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering,” ArXiv160609375 Cs Stat, Feb. 2017, Accessed: Jun. 23, 2020. [Online]. Available: http://arxiv.org/abs/1606.09375.
[32] W. G. Graziano, E. C. Hair, and J. F. Finch, “Competitiveness mediates the link between personality and group performance,” J. Pers. Soc. Psychol., vol. 73, no. 6, pp. 1394–1408, 1997, doi: 10.1037/0022-3514.73.6.1394.
[33] G. A. Van Kleef, A. C. Homan, B. Beersma, and D. van Knippenberg, “On Angry Leaders and Agreeable Followers: How Leaders’ Emotions and Followers’ Personalities Shape Motivation and Team Performance,” Psychol. Sci., vol. 21, no. 12, pp. 1827–1834, Dec. 2010, doi: 10.1177/0956797610387438.
[34] R. E. de Vries, B. van den Hooff, and J. A. de Ridder, “Explaining Knowledge Sharing: The Role of Team Communication Styles, Job Satisfaction, and Performance Beliefs,” Commun. Res., vol. 33, no. 2, pp. 115–135, Apr. 2006, doi: 10.1177/0093650205285366.
[35] T. Sy, S. Côté, and R. Saavedra, “The Contagious Leader: Impact of the Leader’s Mood on the Mood of Group Members, Group Affective Tone, and Group Processes,” J. Appl. Psychol., vol. 90, pp. 295–305, Apr. 2005, doi: 10.1037/0021-9010.90.2.295.
[36] J. M. P. Gevers and M. A. G. Peeters, “A pleasure working together? The effects of dissimilarity in team member conscientiousness on team temporal processes and individual satisfaction,” J. Organ. Behav., vol. 30, no. 3, pp. 379–400, 2009, doi: 10.1002/job.544.
[37] O. Turel and Y. (Jenny) Zhang, “Does virtual team composition matter? Trait and problem-solving configuration effects on team performance,” Behav. Inf. Technol., vol. 29, no. 4, pp. 363–375, Jul. 2010, doi: 10.1080/01449291003752922.
[38] Y.-S. Lin and C.-C. Lee, “Predicting Performance Outcome with a Conversational Graph Convolutional Network for Small Group Interactions,” in ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), May 2020, pp. 8044–8048, doi: 10.1109/ICASSP40776.2020.9053308.