This disclosure describes techniques to perform disambiguation of uncertain or inaccurate user inputs to a computing device. Uncertain input can occur, for example, by an inadvertent touching of the touchscreen of a smartphone in a user’s pocket, inadvertent detection of gesture, etc. Current devices often interpret such input incorrectly, and respond in a manner that is unexpected for the user. Techniques disclosed herein determine a user intent based on the detected gesture. Machine learning is used to disambiguate user-intended input from inadvertent input. The framework described herein detects input and triggers an analyzer. The analyzer receives data from various sources, based on user permission to access such data. The data can include a state of the device, device configuration, user context, sensor data, history, etc. Results from the analyzer are provided to a resolver that determines the user intent and performs an appropriate action, e.g., triggering an application on the device.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.