Contextual privacy engine for notifications
View Patent ↗In an example, there is disclosed a computing apparatus, including a user notification interface; a context interface; and one or more logic elements forming a contextual privacy engine operable to: receive a notification; receive a context via the context interface; apply the context to the notification via a notification rule; and take an action via the user notification interface based at least in part on the applying. The contextual privacy engine may also be operable to mathematically incorporate user feedback into the notification rule. There is also described a method of providing a contextual privacy engine, and one or more computer-readable storage mediums having stored thereon executable instructions for providing a contextual privacy engine.
1. A computing apparatus, comprising:
a hardware platform comprising at least a processor, a memory, and a user notification interface;
a context interface;
a feedback interface; and
one or more logic elements configured to provide a contextual privacy engine on the hardware platform, the contextual privacy engine configured to:
receive a notification;
receive a context via the context interface;
apply the context to the notification via a notification rule;
receive a feedback via the feedback interface;
receive an existing confidence score for the notification rule (score old );
update the existing confidence score to a new confidence score (score new ) based at least in part on the feedback, wherein updating the confidence score comprises an operation of the form score new (A i )=min(score old (A i )+f1*f2*Δ+c,1+t), wherein A represents a set of a plurality of possible actions, i represents an index of the action, t represents a threshold value, f1 represents a fraction of “true” feedback received, f2 represents a percentage of active rules with A i , as the preferred action, Δ represents a difference between t and score old (A i ), and c is a rate-of-growth constant, wherein each of the inputs of the new confidence score is stored in a memory or received as an input; and
take an action via the user notification interface based at least in part on the new confidence score (score new ).
2. The computing apparatus of claim 1 , wherein the action is selected from the group consisting of notify now, notify later, dismiss notification, mute notification, subdue notification, encrypt notification, encode notification, and scramble notification.
3. The computing apparatus of claim 1 , wherein the context comprises an input selected from the group consisting of location, motion, speed, other users present, other users' identities, time, calendar events, third-party program operation, third-party program inputs, input method, incoming application programming interface (API) data, outgoing API data, environmental factors, biometric data, user recognition, authentication strength, alternative authentication, and security tokens.
4. The computing apparatus of claim 1 , wherein applying the context to the notification via the notification rule comprises examining message semantics.
5. The computing apparatus of claim 4 , wherein message semantics comprise at least a message subject matter type.
6. The computing apparatus of claim 1 , wherein the contextual privacy engine comprises a natural language processing engine configured to perform semantic analysis of message content.
7. The computing apparatus of claim 1 , wherein the contextual privacy engine comprises an image processing engine configured to receive an optical input and perform object recognition on the optical input.
8. The computing apparatus of claim 1 , wherein the contextual privacy engine comprises an application programming interface (API) configured to receive third-party program data.
9. The computing apparatus of claim 8 , wherein the API is a native operating system API.
10. The computing apparatus of claim 1 , wherein the contextual privacy engine is configured to build a context vector configured to represent an overall current context.
11. The computing apparatus of claim 10 , wherein the overall current context includes both contextual data and message semantics.
12. One or more tangible, non-transitory computer-readable storage mediums having stored thereon executable instructions for providing a contextual privacy engine, the contextual privacy engine configured to:
receive a notification;
receive a context;
apply the context to the notification via a notification rule;
receive a feedback;
receive an existing confidence score for the notification rule (score old );
update the existing confidence score to a new confidence score (score new ) based at least in part on the feedback, wherein updating the confidence score comprises an operation of the form score new (A i )=min(score old (A i )+f1*f2*Δ+c, 1+t), wherein A represents a set of a plurality of possible actions, i represents an index of the action, t represents a threshold value, f1 represents a fraction of “true” feedback received, f2 represents a percentage of active rules with A i as the preferred action, Δ represents a difference between t and score old (A i ), and c is a rate-of-growth constant, wherein each of the inputs of the new confidence score is stored in a memory or received as an input; and
take an action via a user notification interface based at least in part on the new confidence score (score new ).
13. The one or more tangible, non-transitory computer-readable storage mediums of claim 12 , wherein the action is selected from the group consisting of notify now, notify later, dismiss notification, mute notification, subdue notification, encrypt notification, encode notification, and scramble notification.
14. The one or more tangible, non-transitory computer-readable storage mediums of claim 12 , wherein the context comprises an input selected from the group consisting of location, motion, speed, other users present, other users' identities, time, calendar events, third-party program operation, third-party program inputs, input method, incoming application programming interface (API) data, outgoing API data, environmental factors, biometric data, user recognition, authentication strength, alternative authentication, and security tokens.
15. The one or more tangible, non-transitory computer-readable storage mediums of claim 12 , wherein applying the context to the notification via the notification rule comprises examining message semantics.
16. The one or more tangible, non-transitory computer-readable storage mediums of claim 12 , wherein the contextual privacy engine comprises a natural language processing engine configured to perform semantic analysis of message content.
17. The one or more tangible, non-transitory computer-readable storage mediums of claim 12 , wherein the contextual privacy engine comprises an image processing engine configured to receive an optical input and perform object recognition on the optical input.
18. The one or more tangible, non-transitory computer-readable storage mediums of claim 12 , wherein the contextual privacy engine comprises an application programming interface (API) configured to receive third-party program data.
19. The one or more tangible, non-transitory computer-readable storage mediums of claim 12 , wherein the contextual privacy engine is configured to build a context vector configured to represent an overall current context.
20. A computer-implemented method of providing a contextual privacy engine, comprising:
receiving a notification;
receiving a context;
applying the context to the notification via a notification rule;
receiving a feedback;
receiving an existing confidence score for the notification rule (score old );
updating the existing confidence score to a new confidence score (score new ) based at least in part on the feedback, wherein updating the confidence score comprises an operation of the form score new (A i )=min(score old (A i )+f1*f2*Δ+c,1+t), wherein A represents a set of a plurality of possible actions, i represents an index of the action, t represents a threshold value, f1 represents a fraction of “true” feedback received, f2 represents a percentage of active rules with A i as the preferred action, Δ represents a difference between t and score old (A i ), and c is a rate-of-growth constant, wherein the inputs of the new confidence score are retrieved from a memory or received as an input; and
taking an action via a user notification interface based at least in part on the new confidence score (score new ).
21. The computer-implemented method of claim 20 , wherein the action is selected from the group consisting of notify now, notify later, dismiss notification, mute notification, subdue notification, encrypt notification, encode notification, and scramble notification.
22. The computer-implemented method of claim 20 , wherein the context comprises an input selected from the group consisting of location, motion, speed, other users present, other users' identities, time, calendar events, third-party program operation, third-party program inputs, input method, incoming application programming interface (API) data, outgoing API data, environmental factors, biometric data, user recognition, authentication strength, alternative authentication, and security tokens.
23. The computer-implemented method of claim 20 , wherein applying the context to the notification via the notification rule comprises examining message semantics.
24. The computer-implemented method of claim 23 , wherein message semantics comprise at least a message subject matter type.
25. The computer-implemented method of claim 20 , wherein the contextual privacy engine comprises an image processing engine configured to receive an optical input and perform object recognition on the optical input.