Dynamic linguistic descriptions of time series applied to self-track the physical activity

Self-tracking our physical activity allows us to acquire a better self-knowledge about our physical, and even mental health condition. Currently, the description of the time series provided by sensors is done by means of graphics and tables, that are hard to interpret by non-expert humans. Here, we present a computational application that dynamically describes in natural language the physical activity. The final reports are adapted to the everyday language and user's needs. The application highlights the relevant information obtained at different levels of temporal detail. We have included experimental results that demonstrate the flexibility and applicability of the new tool.