Techniques for graph based natural language processing

1. A computer-implemented method, comprising:

receiving a natural language user request at a client device;

relating the natural language user request to one or more local objects in a local object graph stored on the client device;

for each of the one or more local objects, determining a confidence level that the respective local objects correspond to a natural language meaning of the user request;

determining a low-confidence local object from among the one or more local objects, wherein the confidence level for the low-confidence local object is below a predetermined confidence threshold;

receiving, from a network system, a remote object to replace the low-confidence local object; and

executing the natural language user request based on the remote object.

2. The method of claim 1 , wherein the object graph comprises token mappings for objects within the object graph, the token mappings based on data extracted from a plurality of interactions by a plurality of users of a network system, wherein the one or more objects are matched with the natural language user request based on the token mappings.

3. The method of claim 1 , wherein the confidence level for is further based on a confidence that the respective local object corresponds to an intent of the user request.

4. The method of claim 1 , further comprising transmitting a network query to the network system, the network query requesting that the network system match the natural language user request against a network system object graph, wherein the remote object is received in response to the network query and is represented in the network system object graph.

5. The method of claim 1 , the local objects stored on the client device prior to receiving the natural language user request.

6. The method of claim 1 , the local objects comprising at least one object not represented in an object graph stored on a network system.

7. The method of claim 1 , at least one of the local objects retrieved or generated based on local data stored on the client device.

8. An apparatus, comprising:

a processor circuit on a mobile device;

client service logic operative on the processor circuit to receive a natural language user request;

local object relation logic operative on the processor circuit to relate the natural language user request to one or more local objects in a local object graph stored on the client device, to, for each of the one or more local objects, determine a confidence level that the respective local objects correspond to a natural language meaning of the user request, and to determine a low-confidence local object from among the one or more local objects, wherein the confidence level for the low-confidence local object is below a predetermined confidence threshold;

remote object relation logic operative on the processor circuit to receive, from a network system, a remote object to replace the low-confidence local object; and

user service logic operative on the processor circuit to execute the natural language user request based on the remote object.

9. The apparatus of claim 8 , wherein the object graph comprises token mappings for objects within the object graph, the token mappings based on data extracted from a plurality of interactions by a plurality of users of a network system, wherein the one or more objects are matched with the natural language user request based on the token mappings.

10. The apparatus of claim 8 , wherein the confidence level for is further based on a confidence that the respective local object corresponds to an intent of the user request.

11. The apparatus of claim 8 , the remote object relation logic further configured to transmit a network query to the network system, the network query requesting that the network system match the natural language user request against a network system object graph, wherein the remote object is received in response to the network query and is represented in the network system object graph.

12. The apparatus of claim 8 , the local objects stored on the client device prior to receiving the natural language user request.

13. The apparatus of claim 8 , the local objects comprising at least one object not represented in an object graph stored on a network system.

14. The apparatus of claim 8 , at least one of the local objects retrieved or generated based on local data stored on the client device.

15. At least one non-transitory computer-readable storage medium comprising instructions that, when executed, cause a system to:

receive a natural language user request at a client device;

relate the natural language user request to one or more local objects in a local object graph stored on the client device;

for each of the one or more local objects, determine a confidence level that the respective local objects correspond to a natural language meaning of the user request;

determine a low-confidence local object from among the one or more local objects, wherein the confidence level for the low-confidence local object is below a predetermined confidence threshold;

receive, from a network system, a remote object to replace the low-confidence local object; and

execute the natural language user request based on the remote object.

16. The medium of claim 15 , wherein the object graph comprises token mappings for objects within the object graph, the token mappings based on data extracted from a plurality of interactions by a plurality of users of a network system, wherein the one or more objects are matched with the natural language user request based on the token mappings.

17. The medium of claim 15 , wherein the confidence level for is further based on a confidence that the respective local object corresponds to an intent of the user request.

18. The medium of claim 15 , further storing instructions for transmitting a network query to the network system, the network query requesting that the network system match the natural language user request against a network system object graph, wherein the remote object is received in response to the network query and is represented in the network system object graph.

19. The medium of claim 15 , the local objects stored on the client device prior to receiving the natural language user request.

20. The medium of claim 15 , the local objects comprising at least one object not represented in an object graph stored on a network system.