Methods for enhancing digital search results based on task-oriented user activity
View Patent ↗Methods for using task-related information to enhance digital searching are provided. A task-oriented user activity system maintains task-related information about resources accessed by a user and current user task. This task-related information is used to enhance search results by filtering and ranking results to increase relevance with respect to a user's current task. The task-related information can also be used to include task-related metadata in search engine index, e.g., by storing the metadata in the index or by storing it in resources which are subsequently indexed. Task-related information can also be used to enhance search results by enhancing search queries to include task-related search criteria.
1. A computer-implemented method for digital search with task-enhanced search results, the method comprising:
generating a search query based on interaction with a user;
sending the search query to a search engine for an initial search;
receiving initial search results from the search engine;
predicting a current task being performed by the user at the time of the initial search, the current task being an action that the user is intending to perform at the time the search query is generated, wherein the current task corresponds to the search query without being specified by the search query, wherein predicting the current task includes associating with each of a plurality of tasks a probability that the current task is a particular task from among the plurality of tasks, wherein the associating uses past event records and associated task identifiers stored in a database, the associated task identifiers identifying the plurality of tasks from which the current task is predicted, and wherein the predicted current task is the particular task having a highest probability;
computing task-related information from the predicted current task;
filtering and ranking the initial search results based on the computed task-related information to produce enhanced search results; and
presenting the enhanced search results to the user.
2. The method of claim 1 further comprising, prior to sending the search query to the search engine, transforming the search query to include task-related search criteria.
3. The method of claim 1 wherein the search engine is a desktop search application.
4. The method of claim 1 wherein the search engine is an internet search engine.
5. The method of claim 1 wherein the task-related information includes resources associated with the current task.
6. The method of claim 1 wherein the task-related information includes resource access information.
7. The method of claim 1 wherein the ranking comprises calculating for each resource referenced in the initial search results a likelihood that the resource is associated with the current task being performed by the user.
8. The method of claim 1 wherein the ranking comprises calculating for each resource referenced in the initial search results similarity to resources recently accessed by the user.
9. The method of claim 1 wherein the ranking comprises calculating for each resource referenced in the initial search results a similarity to resources frequently accessed by the user.
10. The method of claim 1 wherein the ranking comprises calculating for each resource referenced in the initial search results a likelihood that the resource has a dataflow connected to resources that are part of the user's likely project.
11. The method of claim 1 wherein the filtering comprises eliminating a resource referenced in the initial search results if a likelihood that the resource is associated with the current task being performed by the user is below a predetermined threshold.
12. The method of claim 1 further comprising
calculating a similarity between keywords associated with the current task and words associated with a resource referenced in the initial search results.
13. The method of claim 12 further comprising
deriving the words associated with the resource referenced in the initial search results by extracting keywords from the resource.
14. The method of claim 12 further comprising
deriving the words associated with the resource referenced in the initial search results by extracting keywords from metadata associated with the resource.
15. The method of claim 12 further comprising
deriving keywords associated with the current task by combining term frequency vectors extracted from all of the resources associated with the task.
16. The method of claim 1 wherein predicting a current task being performed by a user includes using a task-oriented user activity system.
17. A digital search system, comprising:
a search engine that receives a search query for an initial search based on interaction with a user and provides initial search results;
a database containing past event records and associated task identifiers, wherein task-related information is computed from a predicted current task that is predicted as a current task being performed by the user at the time of the initial search and from past event records and associated task identifiers, the current task being an action that the user is intending to perform at the time the search query is generated, wherein the current task corresponds to the search query without being specified by the search query, wherein the predicted current task is predicted in a process that includes associating with each of a plurality of tasks a probability that the current task is a particular task from among the plurality of tasks, wherein the associating uses the past event records and the associated task identifiers stored in the database, the associated task identifiers identifying the plurality of tasks from which the current task is predicted, and wherein the predicted current task is the particular task having a highest probability; and
a task related filtering and ranking component that filters and ranks the initial search results based on the computed task-related information to produce enhanced search results that are presented to the user.
18. The digital search system of claim 17 wherein the search engine is a desktop search application.
19. The digital search system of claim 17 wherein the search engine is an internet search engine.
20. The digital search system of claim 17 wherein the task related filtering and ranking component calculates for each resource referenced in the initial search results a likelihood that the resource is associated with the current task being performed by the user.
21. The digital search system of claim 17 wherein the task related filtering and ranking component calculates for each resource referenced in the initial search results similarity to resources recently accessed by the user.
22. The digital search system of claim 17 wherein the task related filtering and ranking component calculates for each resource referenced in the initial search results a similarity to resources frequently accessed by the user.
23. The digital search system of claim 17 wherein the task related filtering and ranking component calculates for each resource referenced in the initial search results a likelihood that the resource has a dataflow connected to resources that are part of the user's likely project.
24. The digital search system of claim 17 wherein the task related filtering and ranking component eliminates a resource referenced in the initial search results if a likelihood that the resource is associated with the current task being performed by the user is below a predetermined threshold.
25. The digital search system of claim 17 wherein the task related filtering and ranking component calculates a similarity between keywords associated with the current task and words associated with a resource referenced in the initial search results.
26. The digital search system of claim 25 wherein the task related filtering and ranking component derives the words associated with the resource referenced in the initial search results by extracting keywords from the resource.
27. The digital search system of claim 25 wherein the task related filtering and ranking component derives the words associated with the resource referenced in the initial search results by extracting keywords from metadata associated with the resource.
28. The digital search system of claim 25 wherein the task related filtering and ranking component derives keywords associated with the current task by combining term frequency vectors extracted from all of the resources associated with the task.
29. The digital search system of claim 17 , further comprising:
a task-oriented user activity system that receives initial search results from the search engine and predicts the current task being performed by the user.
30. A digital search system, comprising:
means for generating a search query based on interaction with a user;
means for sending the search query to a search engine for an initial search;
means for receiving initial search results from the search engine;
means for predicting a current task being performed by a user at the time of the initial search, the current task being an action that the user is intending to perform at the time the search query is generated, wherein the current task corresponds to the search query without being specified by the search query, wherein predicting the current task includes associating with each of a plurality of tasks a probability that the current task is a particular task from among the plurality of tasks, wherein the associating uses past event records and associated task identifiers stored in a database, the associated task identifiers identifying the plurality of tasks from which the current task is predicted, and wherein the predicted current task is the particular task having a highest probability;
means for computing task-related information from the predicted current task;
means for filtering and ranking the initial search results based on the computed task-related information to produce enhanced search results; and
means for presenting the enhanced search results to the user.
31. The digital search system of claim 30 wherein means for predicting a current task being performed by a user includes a task-oriented user activity system that receives initial search results from the search engine and predicts the current task being performed by the user.