Dennis Küster

Damping sentiment analysis in online communication: discussions, monologs and dialogs

Sentiment analysis programs are now sometimes used to detect patterns of sentiment use over time in online communication and to help automated systems interact better with users. Nevertheless, it seems that no previous published study has assessed whether the position of individual texts within on-going communication can be exploited to help detect their sentiments. This article assesses apparent sentiment anomalies in on-going communication --- texts assigned significantly different sentiment strength to the average of previous texts --- to see whether their classification can be improved. The results suggest that a damping procedure to reduce sudden large changes in sentiment can improve classification accuracy but that the optimal procedure will depend on the type of texts processed.

Empathic robotic tutors for personalised learning : A multidisciplinary approach

Within any learning process, the formation of a socio-emotional relationship between learner and teacher is paramount to facilitating a good learning experience. The ability to form this relationship may come naturally to an attentive teacher; but how do we endow an unemotional robot with this ability? In this paper, we extend upon insights from the literature to include tools from user-centered design (UCD) and analyses of human-human interaction (HHI) as the basis of a multidisciplinary approach in the development of an empathic robotic tutor. We discuss the lessons learned in respect to design principles with the aim of personalised learning with empathic robotic tutors.

Mixing implicit and explicit probes: finding a ground truth for engagement in social human-robot interactions

In our work we explore the development of a computational model capable of automatically detecting engagement in social human-robot interactions from real-time sensory and contextual input. However, to train the model we need to establish ground truths of engagement from a large corpus of data collected from a study involving task and social-task engagement. Here, we intend to advance the current state-ofthe-art by reducing the need for unreliable post-experiment questionnaires and costly time-consuming annotation with the novel introduction of implicit probes. A non-intrusive, pervasive and embedded method of collecting informative data at different stages of an interaction.

Applying a Text-Based Affective Dialogue System in Psychological Research: Case Studies on the Effects of System Behaviour, Interaction Context and Social Exclusion

This article presents two studies conducted with an affective dialogue system in which text-based system–user communication was used to model, generate and present different affective and social interaction scenarios. We specifically investigated the influence of interaction context and roles assigned to the system and the participants, as well as the impact of pre-structured social interaction patterns that were modelled to mimic aspects of “social exclusion” scenarios. The results of the first study demonstrate that both the social context of the interaction and the roles assigned to the system influence the system evaluation, interaction patterns, textual expressions of affective states, as well as emotional self-reports. The results observed for the second study show the system’s ability to partially exclude a participant from a triadic conversation without triggering significantly different affective reactions or a more negative system evaluation. The experimental evidence provides insights on the perception, modelling and generation of affective and social cues in artificial systems that can be realized in different modalities, including the text modality, thus delivering valuable input for applying affective dialogue systems as tools for studying affect and social aspects in online communication.

A Review of Dynamic Datasets for Facial Expression Research

Temporal dynamics have been increasingly recognized as an important component of facial expressions. With the need for appropriate stimuli in research and application, a range of databases of dynamic facial stimuli has been developed. The present article reviews the existing corpora and describes the key dimensions and properties of the available sets. This includes a discussion of conceptual features in terms of thematic issues in dataset construction as well as practical features which are of applied interest to stimulus usage. To identify the most influential sets, we further examine their citation rates and usage frequencies in existing studies. General limitations and implications for emotion research are noted and future directions for stimulus generation are outlined.

Engagement Perception and Generation for Social Robots and Virtual Agents

Technology is the future , woven into every aspect of our lives, but how are we to interact with all this technology and what happens when problems arise? Artificial agents, such as virtual characters and social robots could offer a realistic solution to help facilitate interactions between humans and machines—if only these agents were better equipped and more informed to hold up their end of an interaction. People and machines can interact to do things together, but in order to get the most out of every interaction, the agent must to be able to make reasonable judgements regarding your intent and goals for the interaction. We explore the concept of engagement from the different perspectives of the human and the agent. More specifically, we study how the agent perceives the engagement state of the other interactant, and how it generates its own representation of engaging behaviour. In this chapter, we discuss the different stages and components of engagement that have been suggested in the literature from the applied perspective of a case study of engagement for social robotics, as well as in the context of another study that was focused on gaze-related engagement with virtual characters.

Social Effects of Tears and Small Pupils Are Mediated by Felt Sadness: An Evolutionary View:

Small pupils elicit empathic socioemotional responses comparable to those found for emotional tears. This might be understood in an evolutionary context. Intense emotional tearing increases tear film volume and disturbs tear layer uniformity, resulting in blurry vision. A constriction of the pupils may help to mitigate this handicap, which in turn may have resulted in a perceptual association of both signals. However, direct empirical evidence for a role of pupil size in tearful emotional crying is still lacking. The present study examined socioemotional responses to different pupil sizes, combined with the presence (absence) of digitally added tears superimposed upon expressively neutral faces. Data from 50 subjects showed significant effects of observing digitally added tears in avatars, replicating previous findings for increased perceived sadness elicited by tearful photographs. No significant interactions were found between tears and pupil size. However, small pupils likewise elicited a significantly greater wish to help in observers. Further analysis showed a significant serial mediation of the effects of tears on perceived wish to help via perceived and then felt sadness. For pupil size, only felt sadness emerged as a significant mediator of the wish to help. These findings support the notion that pupil constriction in the context of intense sadness may function to counteract blurry vision. Pupil size, like emotional tears, appears to have acquired value as a social signal in this context.

Measuring Emotions Online: Expression and Physiology

Cyberemotions refer to emotions in networks that are a complex function of emotional states in individuals. Thus, measuring cyberemotions frequently involves attempts to estimate emotional states in individuals. Yet, this is not easy, as emotions in individuals are characterized by limited cohesion of the components of response, such as expression in the face, voice, and body, central and peripheral physiological changes, changes in action readiness, as well as subjective experience. There is no gold standard that would identify any of these components by a single criterion. In consequence, modern experimental emotion research has focused on multi-modal assessment of emotions. Different components are targeted at identifying the valence of responses, or the intensity. We will describe paradigms that are particularly tailored for research in the context of cyberemotions and illustrate these with concrete examples of data recorded in our laboratory.

Students' Normative Perspectives on Classroom Robots

As robots are becoming increasingly common in society and education, it is expected that autonomous and socially adaptive classroom robots may eventually be given responsible roles in primary education. In this paper, we present the results of a questionnaire study carried out with students enrolled in compulsory education in three European countries. The study aimed to explore students’ normative perspectives on classroom robots pertaining to roles and responsibilities, student-robot relationships, and perceptive and emotional capabilities in robots. The results suggest that, although students are generally positive toward the existence of classroom robots, certain aspects are deemed more acceptable than others.

Boxing the face: A comparison of dynamic facial databases used in facial analysis and animation

Facial animation is a difficult task that is based on an approximation of subtle facial movements [Trutoiu et al. 2014], and that needs to be well grounded in real life dynamic facial behaviour to be convincing. Yet while the endpoints of expressions in still images can be defined relatively precisely using the Facial Action Coding System FACS [Ekman et al. 2002], the design of facial dynamics requires additional highresolution data (e.g., [Trutoiu et al. 2014]). This is particularly the case for the creation of more naturalistic expressions, i.e., everyday patterns of “partial” and “mixed” movements that depart from simplistic assumptions of omnipresent stereotypic expressions of “basic” patterns of Action Units (AUs). However, for animation designers who do not have the resources to elicit, record, and validate such expressions, the question arises which of the extant and freely available dynamic facial databases might best serve this purpose. One of the most important steps in this decision process is the selection of technically adequate databases that offer sufficient resolution of the face and expressions to allow adequate modelling.