Modeling Empathy and Distress in Reaction to News Stories

Abstract

Computational detection and understanding of empathy is an important factor in advancing human-computer interaction. Yet to date, textbased empathy prediction has the following major limitations: It underestimates the psychological complexity of the phenomenon, adheres to a weak notion of ground truth where empathic states are ascribed by third parties, and lacks a shared corpus. In contrast, this contribution presents the first publicly available gold standard for empathy prediction. It is constructed using a novel annotation methodology which reliably captures empathy assessments by the writer of a statement using multiitem scales. This is also the first computational work distinguishing between multiple forms of empathy, empathic concern, and personal distress, as recognized throughout psychology. Finally, we present experimental results for three different predictive models, of which a CNN performs the best.

Design and Evaluation of an Adaptive Empathy Learning Tool

“…The design and implementation of ITS and CSCL to build adaptive learning tools is a rather complex endeavor that must rely on expertise from the fields of computer science (i.e., development of feedback algorithms), human-computer interaction (i.e., design of the interface) and educational technology (i.e., integration into the learning process). Therefore, we aim to address this research gap and rigorously design an empathy learning tool based on educational theory through the application of recent developments in NLP and ML (e.g., [28], [29]), in which empathy detection has been a growing research approach to identify and model empathetic structures of a given text in real time [11], [12]. The potential of empathy detection has been investigated in different domains but not leveraged for individual tutoring or feedback in a student's learning progress [12].…”

“…Therefore, we aim to address this research gap and rigorously design an empathy learning tool based on educational theory through the application of recent developments in NLP and ML (e.g., [28], [29]), in which empathy detection has been a growing research approach to identify and model empathetic structures of a given text in real time [11], [12]. The potential of empathy detection has been investigated in different domains but not leveraged for individual tutoring or feedback in a student's learning progress [12].…”

“…However, novel technologyenhanced pedagogical scenarios based on recent advances in Natural Language Processing (NLP) or Machine Learning (ML) that allow new forms of human-computer interaction to support learners in learning empathetic interaction through adaptive tutoring fall rather short in literature. A possibility to provide adaptive empathy feedback on natural language is the field of empathy detection form Computational Linguistics [12]. Empathy detection has been a growing research approach to identify and model empathetic structures and phrases of a given text in real time, which could be leveraged to provide students with individual feedback, e.g., on peer reviews on business models or team conversation logs [11], [12].…”

“…A possibility to provide adaptive empathy feedback on natural language is the field of empathy detection form Computational Linguistics [12]. Empathy detection has been a growing research approach to identify and model empathetic structures and phrases of a given text in real time, which could be leveraged to provide students with individual feedback, e.g., on peer reviews on business models or team conversation logs [11], [12]. However, despite the vast amount of studies, current literature falls short of providing an approach with principles and proof on how to design an adaptive and intelligent learning tool that helps students to learn how to react to other students' perspectives with intelligent feedback.…”

Condolence and Empathy in Online Communities

“…Distressed individuals have found empathetic condolences more supportive than sympathetic messages (Davidowitz and Myrick, 1984 (Basch, 1983;Cuff et al, 2016) and the limited computational work employing empathy has largely focused only on mirroring emotional state as a way of empathizing (Collins, 2014;Litvak et al, 2016;Fung et al, 2016;Khanpour et al, 2017). More recently Abdul-Mageed et al (2017) and Buechel et al (2018) have gone beyond these simple models to develop and use a corpus for distress and empathy in reactions to news stories. These works adopt a broader definition drawn from multiple sources of empathy which mixes empathy with related concepts of compassion, altruism, and prosocial behavior.…”