Emotion Prediction in Real-Life Scenarios: On the Way to the BIRAFFE3 Dataset

Despite over 20 years of research in affective computing, emotion prediction models that would be useful in real-life out-of-the-lab scenarios such as health care or intelligent assistants have still not been developed. The identification of the fundamental problems behind this concern led to the initiation of the BIRAFFE series of experiments, whose main goal is to develop a set of techniques, tools and good practices to introduce personalized context-based emotion processing modules in intelligent systems/assistants. The aim of this work is to present the work-in-progress concept of the third experiment in the BIRAFFE series and discuss the results of the pilot study. After all conclusions have been drawn up, actual study will be carried out, and then the collected data will be processed and made available under the creative commons license as BIRAFFE3 dataset.