Predicting intent of a search for a particular context

1. A method comprising:

receiving, by a computing system, a search query from a computing device of a user, the search query including one or more terms;

obtaining, by the computing system, contextual information associated with the computing device of the user, wherein the contextual information is in addition to the one or more terms of the search query;

selecting, by the computing system, a subset of the obtained contextual information that indicates a particular current context associated with the computing device, wherein at least the subset of the obtained contextual information excludes information from a search history of the user;

determining, by the computing system, a task that the user is likely to perform given the particular current context, wherein the task is a predicted task that is specific to the particular current context associated with the computing device, and wherein determining the task that the user is likely to perform given the particular current context comprises:

processing the subset of the contextual information, using a machine learning model that is based on past user-initiated actions performed by a group of computing devices of respective users, to determine the task;

receiving, by the computing system, search results obtained responsive to the one or more terms of the search query; and

subsequent to receiving the search results obtained responsive to the one or more terms of the search query:

adjusting, by the computing system, based on the determined task that the user is likely to perform given the particular current context, at least a particular portion of the search results obtained responsive to the one or more terms of the search query; and

sending, by the computing system, to the computing device of the user, an indication of the adjusted search results.

2. The method of claim 1 , wherein selecting the subset of the obtained contextual information that indicates a particular current context associated with the computing device further comprises:

defining, based on the contextual information associated with the computing device of the user and the past user-initiated actions performed by the group of computing devices, the particular current context associated with the computing device, wherein the particular current context associated with the computing device is defined based on relevance to the search query.

3. The method of claim 1 , further comprising:

receiving, by the computing system, log data indicative of user inputs received by the group of computing devices; and

determining, based on the log data, the past user-initiated actions performed by the group of computing devices.

4. The method of claim 3 , wherein:

the search results include a plurality of search features, each search feature being associated with curated content; and

at least the particular portion of the search results comprises a particular search feature from the plurality of search features.

5. The method of claim 4 , wherein the past user-initiated actions performed by the group of computing devices include user-initiated actions performed by the group of computing devices based on receiving user inputs directed to the particular search feature.

6. The method of claim 5 , wherein processing the subset of the contextual information, using a machine learning model that is based on past user-initiated actions performed by a group of computing devices of respective users, to determine the task that the user is likely to perform given the particular current context comprises:

identifying, by the computing system and based on the particular current context associated with the computing device, the particular search feature prior to receiving the search results obtained responsive to the one or more terms of the search query.

7. The method of claim 5 , wherein adjusting at least the particular portion of the search results obtained responsive to the search query further comprises:

identifying, by the computing system, based on the determined task that the user is likely to perform given the particular current context, the particular search feature; and

increasing a respective ranking of the particular search feature relative to respective rankings of other search features from the plurality of search features.

8. The method of claim 3 , wherein the log data includes at least one of application usage data for the group of computing devices for a plurality of different contexts or search feature data for the group of computing devices for the plurality of different contexts.

9. A system comprising:

one or more processors; and

memory storing instructions that, when executed by the one or more processors, cause the one or more processors to:

receive a search query from a computing device of a user, the search query including one or more terms;

obtain contextual information associated with the computing device of the user, wherein the contextual information is in addition to the one or more terms of the search query;

select a subset of the obtained contextual information that indicates a particular current context associated with the computing device, wherein at least the subset of the obtained contextual information excludes information from a search history of the user;

determine a task that the user is likely to perform given the particular current context, wherein the task is a predicted task that is specific to the particular current context associated with the computing device, and wherein determining the task that the user is likely to perform given the particular current context comprises:

processing the subset of the contextual information, using a machine learning model that is based on past user-initiated actions performed by a group of computing devices of respective users, to determine the task;

receive search results obtained responsive to the one or more terms of the search query; and

subsequent to receiving the search results obtained responsive to the one or more terms of the search query:

adjust, based on the determined task that the user is likely to perform given the particular current context, at least a particular portion of the search results obtained responsive to the one or more terms of the search query; and

send, to the computing device of the user, an indication of the adjusted search results.

10. The system of claim 9 , wherein:

the search results include a plurality of search features, each search feature being associated with curated content; and

at least the particular portion of the search results comprises a particular search feature from the plurality of search features.

11. The system of claim 10 , wherein processing the subset of the contextual information, using a machine learning model that is based on past user-initiated actions performed by a group of computing devices of respective users, to determine the task that the user is likely to perform given the particular current context comprises:

identifying, based on the particular current context associated with the computing device, the particular search feature prior to receiving the search results obtained responsive to the one or more terms of the search query.

12. The system of claim 9 , wherein determining the task that the user is likely to perform given the particular current context further comprises:

processing the search query, along with the subset of the contextual information and using the machine learning model, to determine the task.

13. The system of claim 9 , wherein promoting at least the particular portion of the search results obtained responsive to the one or more terms of the search query based on the determined task that the user is likely to perform given the particular current context comprises:

identifying that a first search result and a second search result are each associated with the determined task that the user is likely to perform given the particular current context, wherein the first search result and the second search result are responsive to the search query; and

displaying the first search result and the second search result more prominently than other search results, of the search results, based on identifying that the first search result and the second search result are each associated with the determined task that the user is likely to perform given the particular current context.

14. The system of claim 9 , wherein the contextual information includes at least one of:

location information of the computing device,

movement information of the computing device,

weather information for a location of the computing device,

a time of day, or

a day of week.

15. At least one non-transitory computer-readable medium comprising instructions that, in response to execution of the instructions by one or more processors, cause the one or more processors to perform the following operations:

receiving a search query from a computing device of a user, the search query including one or more terms;

obtaining contextual information associated with the computing device of the user, wherein the contextual information is in addition to the one or more terms of the search query;

selecting a subset of the obtained contextual information that indicates a particular current context associated with the computing device, wherein at least the subset of the obtained contextual information excludes information from a search history of the user;

determining a task that the user is likely to perform given the particular current context, wherein the task is a predicted task that is specific to the particular current context associated with the computing device, and wherein determining the task that the user is likely to perform given the particular current context comprises:

processing the subset of the contextual information, using a machine learning model that is based on past user-initiated actions performed by a group of computing devices of respective users, to determine the task;

receiving search results obtained responsive to the one or more terms of the search query; and

subsequent to receiving the search results obtained responsive to the one or more terms of the search query:

adjusting, based on the determined task that the user is likely to perform given the particular current context, at least a particular portion of the search results obtained responsive to the one or more terms of the search query; and

sending, to the computing device of the user, an indication of the adjusted search results.

16. The non-transitory computer-readable medium of claim 15 , the operations further comprising:

receiving log data indicative of user inputs received by the group of computing devices; and

determining, based on the log data, the past user-initiated actions performed by the group of computing devices.

17. The non-transitory computer-readable medium of claim 15 , wherein:

the search results include a plurality of search features, each search feature being associated with curated content;

at least the particular portion of the search results comprises a particular search feature from the plurality of search features; and

the past user-initiated actions performed by the group of computing devices include user-initiated actions performed by the group of computing devices based on receiving user inputs directed to the particular search feature.

18. The non-transitory computer-readable medium of claim 17 , wherein processing the subset of the contextual information, using a machine learning model that is based on past user-initiated actions performed by a group of computing devices of respective users, to determine the task that the user is likely to perform given the particular current context comprises:

identifying, based on the particular current context associated with the computing device, the particular search feature prior to receiving the search results obtained responsive to the one or more terms of the search query.

19. The non-transitory computer-readable medium of claim 18 , wherein adjusting at least the particular portion of the search results obtained responsive to the search query comprises:

increasing a respective ranking of the particular search feature relative to respective rankings of other search features from the plurality of search features.

20. The non-transitory computer-readable medium of claim 18 , wherein processing the subset of the contextual information, using a machine learning model that is based on past user-initiated actions performed by a group of computing devices of respective users, to determine the task that the user is likely to perform given the particular current context further comprises:

selecting a subset of the past user-initiated actions that are performed by the group of computing devices based on receiving the user inputs directed to the particular search feature,

wherein the subset is selected based on the selected user-initiated actions directed to the particular search feature occurring within one or more previous contexts that correspond to the particular current context and not occurring within one or more different previous contexts that do not correspond to the particular current context; and

processing the selected subset of the past user-initiated actions, using the machine learning model that is based on the past user-initiated actions performed by the group of computing devices of respective users, to determine the task that the user is likely to perform given the particular current context.