Media agent to suggest contextually related media content
First Claim
1. A computer-readable storage medium comprising computer-executable instructions for:
- detecting user input corresponding to a present user context; and
responsive to detecting the user input and independent of whether the user input is associated with an explicit query;
analyzing at least a subset of the user input in view of semantic text and user preferences modeling, the semantic text comprising the at least a subset and previously collected text from a personal media database customized for the user, the previously collected text being semantically related to one or more previous multimedia accesses by the user, the user preferences modeling containing user log records clustered into several preferences clusters based on clusters semantic similarity, each cluster of the clusters represented by a keyword frequency vector, the analyzing further comprising evaluating the at least a subset of the user input based on lexical and syntactical features;
predicting desired access to one or more media files based on the analysis;
retrieving information corresponding to the one or more media files from a media content source, wherein the retrieved information was generated in response to a user context previous and different from the present user context;
displaying the retrieved information as a suggestion to a userthe evaluating the at least subset of the user input in view of linguistic features and user intention modeling, the user intention modeling using the linguistic features of the user input to predict a next action of the user; and
displaying an option to execute the next action as a suggestion to the user.
2 Assignments
0 Petitions
Accused Products
Abstract
The described arrangements and procedures provide an intelligent media agent to autonomously collect semantic multimedia data text descriptions on behalf of a user whenever and wherever the user accesses media content. The media agent analyzes these semantic multimedia data text descriptions in view of user behavior patterns and actions to assist the user in identifying multimedia content and related information that is appropriate to the context within which the user is operating or working. For instance, the media agent detects insertion of text and analyzes the inserted text. Based on the analysis, the agent predicts whether a user intends to access media content. If so, the agent retrieves information corresponding to media content from a media content source and presents the information to a user as a suggestion.
328 Citations
21 Claims
-
1. A computer-readable storage medium comprising computer-executable instructions for:
-
detecting user input corresponding to a present user context; and responsive to detecting the user input and independent of whether the user input is associated with an explicit query; analyzing at least a subset of the user input in view of semantic text and user preferences modeling, the semantic text comprising the at least a subset and previously collected text from a personal media database customized for the user, the previously collected text being semantically related to one or more previous multimedia accesses by the user, the user preferences modeling containing user log records clustered into several preferences clusters based on clusters semantic similarity, each cluster of the clusters represented by a keyword frequency vector, the analyzing further comprising evaluating the at least a subset of the user input based on lexical and syntactical features; predicting desired access to one or more media files based on the analysis; retrieving information corresponding to the one or more media files from a media content source, wherein the retrieved information was generated in response to a user context previous and different from the present user context; displaying the retrieved information as a suggestion to a user the evaluating the at least subset of the user input in view of linguistic features and user intention modeling, the user intention modeling using the linguistic features of the user input to predict a next action of the user; and displaying an option to execute the next action as a suggestion to the user. - View Dependent Claims (2, 3, 4, 5, 6, 7)
-
-
8. A computer-implemented method for retrieving information from a media content source, comprising:
-
detecting user input corresponding to a present user context; and responsive to detecting the user input and independent of whether the user input is associated with an explicit query; analyzing at least a subset of the user input in view of semantic text and user preferences modeling, the semantic text comprising the at least a subset and previously collected text from a personal media database customized for the user, the previously collected text being semantically related to one or more previous multimedia accesses by the user, the user preferences modeling containing user log records clustered into several preferences clusters based on clusters semantic similarity, each cluster of the clusters represented by a keyword frequency vector, the analyzing further comprising evaluating the at least a subset of the user input based on lexical and syntactical features; predicting desired access to one or more media files based on the analysis; retrieving information corresponding to the one or more media files from a media content source based on the analysis, wherein the retrieved information was generated in response to a user context previous and different from the present user context; displaying the retrieved information as a suggestion to a user; the evaluating the at least a subset of the user input in view of linguistic features and user intention modeling, the user intention modeling using the linguistic features of the user input to predict a next action of the user; and displaying an option to execute the next action as a suggestion to the user. - View Dependent Claims (9, 10, 11, 12, 13, 14)
-
-
15. A system comprising at least one processor and a computer-accessible storage medium coupled to the at least one processor, the system configured to:
-
detect user input corresponding to a present user context; and responsive to detecting the user input and independent of whether the user input is associated with an explicit query; analyze at least a subset of the user input in view of semantic text and user preferences modeling, the semantic text comprising the at least a subset and previously collected text from a personal media database customized for the user, the previously collected text being semantically related to one or more previous multimedia accesses by the user, the user preferences modeling containing user log records clustered into several preferences clusters based on clusters semantic similarity, each cluster of the clusters represented by a keyword frequency vector, the analyzing further comprising evaluating the at least a subset of the user input based on lexical and syntactical features; predict desired access to one or more media files based on the analysis; retrieve information corresponding to the one or more media files from a media content source based on the analysis, wherein the retrieved information was generated in response to a user context previous and different from the present user context; display the retrieved information as a suggestion to a user; the evaluating the at least a subset of the user input in view of linguistic features and user intention modeling, the user intention modeling using the linguistic features of the user input to predict a next action of the user; and display an option to execute the next action as a suggestion to the user. - View Dependent Claims (16, 17, 18, 19, 20, 21)
-
Specification