Automatic Storytelling from Wearable Sensor Data for Health and Wellness Applications

Storytelling describes our daily living activities in many ways. It assists us to understand what we have done to advance our health and wellbeing. In this paper, we present our novel approach to generate scripts from events, which are detected from wearable sensor data. First, we use Deep Neural Network (DNN) to recognize semantic concepts such as gesture, activity, and location for generating a chronological sequence of events. Second, we apply a sequence to sequence (SEQ2SEQ) model consisting of two recurrent neural networks (RNNs) to generate human-understandable stories. The results show that our method can improve the performance of script generation (SG) by using SEQ2SEQ with 0.972 BLEU-1 score.