A Data Collection and Annotation Tool for Asynchronous Multimodal Data During Human-Computer Interactions

Supervised Learning is a Machine Learning technique where a predictor is trained using a set of labeled data where each data point has a label associated with it. The technique can be used for Affective Computing where a person's physiological signals are measured and used in conjunction with labels to create a predictor for affective states. However, creating this labeled dataset is labor-intensive and prone to human errors. To alleviate some of the costs of manual labeling, a data annotation tool is needed to create labels that can eventually be combined with physiological data to create a fully labeled dataset. The goal of the tool is to streamline the steps from data collection and data annotation to creating an effective computing agent. A use case and initial development of this tool stems from a socially assistive robotics system setting with older adults who have cognitive impairment and dementia. However, the tool was further developed to be adaptive to the needs of any data annotation. In this paper, we present a data annotation tool that can be adapted to any domain for streamlining the process of data annotation.