It is well known that utterances convey a great deal of information about the speaker in addition to their semantic content. One such type of information consists of cues to the speaker’s personality traits, the most fundamental dimension of variation between humans. Recent work explores the automatic detection of other types of pragmatic variation in text and conversation, such as emotion, deception, speaker charisma, dominance, point of view, subjectivity, opinion and sentiment. Personality affects these other aspects of linguistic production, and thus personality recognition may be useful for these tasks, in addition to many other potential applications. However, to date, there is little work on the automatic recognition of personality traits. This article reports experimental results for recognition of all Big Five personality traits, in both conversation and text, utilising both self and observer ratings of personality. While other work reports classification results, we experiment with classification, regression and ranking models. For each model, we analyse the effect of different feature sets on accuracy. Results show that for some traits, any type of statistical model performs significantly better than the baseline, but ranking models perform best

Abstract. A significant goal in multi-modal virtual agent research is to determine how to vary expressive qualities of a character so that it is perceived in a desired way. The “Big Five ” model of personality offers a potential framework for organizing these expressive variations. In this work, we focus on one parameter in this model – extraversion – and demonstrate how both verbal and non-verbal factors impact its perception. Relevant findings from the psychology literature are summarized. Based on these, an experiment was conducted with a virtual agent that demonstrates how language generation, gesture rate and a set of movement performance parameters can be varied to increase or decrease the perceived extraversion. Each of these factors was shown to be significant. These results offer guidance to agent designers on how best to create specific characters.

Latent Response Models, intelligent agents 2 Students use intelligent tutors and other types of interactive learning environments in a considerable variety of ways. In this thesis, I detail my work to understand, automatically detect, and re-design an intelligent tutoring system to adapt to a behavior I term “gaming the system”. Students who game the system attempt to succeed in the learning environment by exploiting properties of the system rather than by learning the material and trying to use that knowledge to answer correctly. Within this thesis, I present a set of studies aimed towards understanding what effects gaming has on learning, and why students game, using a combination of quantitative classroom observations and machine learning. In the course of these studies, I determine that gaming the system is replicably associated with low learning. I use data from these studies to develop a profile of students who game, showing that gaming students have a consistent pattern of negative affect

Conversation is an essential component of social behaviour, one of the primary means by which humans express emotions, moods, attitudes and personality. Thus a key technical capability for dialogue applications, such as interactive narrative systems (INS), human robot interaction (HRI) and spoken dialogue systems (SDS), is the ability to support natural conversational interaction. However, system utterances in existing systems are typically handcrafted, leading to problems of portability and scalability. We propose a framework for automatically generating language projecting different personality traits based on the ‘Big Five ’ model of personality. We show that our PERSONAGE generator can produce utterances with recognisable personality for all Big Five traits, according to human judges. We also test the ability of PERSONAGE to vary the characters ’ personality in an existing interactive narrative system, showing that some forms of variation can be automatically obtained in a new domain, depending on the level of utterance representation.

Recent work in natural language generation has begun to take linguistic variation into account, developing algorithms that are capable of modifying the system’s linguistic style based either on the user’s linguistic style or other factors, such as personality or politeness. While stylistic control has traditionally relied on handcrafted rules, statistical methods are likely to be needed for generation systems to scale to the production of the large range of variation observed in human dialogues. Previous work on statistical natural language generation (SNLG) has shown that the grammaticality and naturalness of generated utterances can be optimized from data; however these data-driven methods have not been shown to produce stylistic variation that is perceived by humans in the way that the system intended. This paper describes PERSONAGE, a highly parameterizable language generator whose parameters are based on psychological findings about the linguistic reflexes of personality. We present a novel SNLG method which uses parameter estimation models trained on personality-annotated data to predict the generation decisions required to convey any combination of scalar values along the five main dimensions of personality. A human evaluation shows that parameter estimation models produce recognizable stylistic variation along multiple dimensions, on a continuous scale, and without the computational cost incurred by overgeneration techniques. 1.

Conversation is an essential component of social behavior, one of the primary means by which humans express intentions, beliefs, emotions, attitudes and personality. Thus a key technical capability for dialogue-based conversational agents for interactive entertainment, therapeutic, or learning applications is the ability to support natural conversational interaction. To do so, natural language processing is often applied to allow users flexibility in what they say to the system, but the system’s output typically consists of highly handcrafted utterances, designed to portray a particular system personality or linguistic style. Although this approach produces high quality output, it leads to an authoring bottleneck, and makes it difficult to personalize or adapt the interaction. We propose a computational framework for system utterance generation that builds on the “Big Five ” model of personality traits, and findings from psychology on the linguistic reflexes of personality. Our results show that our framework reliably generates utterances that humans perceive as manifesting each extreme of all Big Five traits, suggesting that the linguistic markers identified in previous work in naturally occurring language genres can be used to manifest personality in a controlled discourse situation of a conversational agent. 1

Abstract—This paper investigates how the use of machine learning techniques can significantly predict the three major dimensions of learner’s emotions (pleasure, arousal and dominance) from brainwaves. This study has adopted an experimentation in which participants were exposed to a set of pictures from the International Affective Picture System (IAPS) while their electrical brain activity was recorded with an electroencephalogram (EEG). The pictures were already rated in a previous study via the affective rating system Self-Assessment Manikin (SAM) to assess the three dimensions of pleasure, arousal, and dominance. For each picture, we took the mean of these values for all subjects used in this previous study and associated them to the recorded brainwaves of the participants in our study. Correlation and regression analyses confirmed the hypothesis that brainwave measures could significantly predict emotional dimensions. This can be very useful in the case of impassive, taciturn or disabled learners. Standard classification techniques were used to assess the reliability of the automatic detection of learners ’ three major dimensions from the brainwaves. We discuss the results and the pertinence of such a method to assess learner’s emotions and integrate it into a brainwavesensing