Storytelling Machine by Learning Customer Spatio-Temporal Trajectories
Publication Date: 2016-Jan-19
The IP.com Prior Art Database
Disclosed is a mechanism for automatic storytelling by learning from customer spatio-temporal trajectories collected from mobile devices.
Page 01 of 3
Storytelling Machine by Learning Customer Spatio -
Storytelling is an exciting concept for any business looking to construct linear customer profiles for decision-makers. Current storyteller systems used by social networking applications rely heavily on the user's active inputs such as images and texts . The generated stories are subsets of the user inputs that selectively connect the customer's important events in chorological order. The narratives can hardly provide business intelligence without additional human interpretation of the generated contents
Proposed is an efficient framework based on deep neural network to automatically build personal stories for individual customers from the individual's spatio -temporal trajectories.
The novel framework generates personal stories by combining the daily activity records
with the inferred personal characteristics from advanced Artificial Intelligence (AI) technology (i.e. deep neural network). This framework translates the mobile sensing signals to human interpretable text representations that include actors , plots, actions, and surrounding world. The framework utilizes the signals collected automatically by phone usage, with the purpose of predictive intelligence of customers' behavior .
The main objectives of the system are to: (Figure 1)
• Utilize mobile sensor collected signals from phone usage
• Translate user daily movement activity signals into summarizations of customer
• Perform predictive analysis from the machine generated customer stories
Figure 1: Storytelling Machine by Learning Customer Spatio-Temporal Trajectories
The steps for implementing the proposed solution follow : (Figure 2)
1. Pre-process the data, then design features that can express semantics such as temporal ordering and spatial scenes for each customer
2. Model the customer behavior via deep architecture (e.g., Recurrent Neural
Page 02 of 3
Network) and learn the...