"Towards automated e-counselling system based on counsellors emotion perception"
Emotions are a core semantic component of human communication. Since counsellors are humans we assume that their own state of emotions could affect their intuitional effort when taking decisions concerning their clients. Therefore, the accuracy of detected emotions by counsellors could be doubtful. And this highlights the need for complementing the intuitional effort of counsellors by computational approach. Therefore, ascertaining the efficacy of computational algorithm, there is the need to benchmark with humans. In this paper, we explore empirically, the extent to which counsellors own emotional states influence their perception of emotions expressed in text. This influence is investigated through the level of agreement among counsellors when annotating emotions expressed in students’ personal life’s stories. The result shows strong intra-counsellor annotation agreement of emotions while inter-counsellors annotation agreement was low. Furthermore, the intra-annotation agreement of emotions was found to be strongly correlated to the counsellors’ self-reported emotions. We speculate, based on the findings, that the emotional state of counsellors influences their emotion perception while tracking emotions in text. Based on the results, we discuss the advantages of using an automated e-counselling system for emotion analysis.
KeywordsArtificial Intelligence, Chat-Bot, Psychology, Counselling, Clinical Work, Deep Learning, Machine Learning, Emotions, Technology
ThemesAI and Counselling
Links to Reference
How to contribute.