Media event structure and context identification using short messages
View Patent ↗The present disclosure is descriptive of discovering structure, content, and context of a media event, e.g., a live media event, using real-time discussions that unfold through short messaging services. Generally, a sampling of short messages of a plurality of users is obtained. The sampling of short messages corresponds to a media event. A segment in the media event is identified using the sampling of short messages, and at least one term taken from the sampling of short messages is identified. The at least one term is indicative of a context of the identified segment.
1. A method comprising:
obtaining, using at least one computing device, a sampling of short messages with content posted by a plurality of users, by sampling a collection of short messages with content posted during a broadcast of a media event's content;
identifying, using the at least one computing device and the sampling of short messages, a segment in the media event's content, identification of the segment comprising identifying an onset and an end of the identified segment in the media event's content using levels of short messaging activity detected using the sampling of short messages, the levels comprising a level used in identifying the onset and a different level used in identifying the end;
identifying, using the at least one computing device, at least one term from the content of the sampling of short messages, the at least one term indicative of a context of the identified segment in the media event content; and
annotating, using the at least one computing device, the media event's content using the at least one term identified from the content of the sampling of short messages.
2. The method of claim 1 , the annotating the media event's content further comprising:
annotating, using the at least one computing device, the segment identified in the media event's content using the at least one term identified from the content of the sampling of short messages.
3. The method of claim 1 , further comprising:
selecting, using the at least one computing device, the sampling of short messages from a collection of short messages, the selecting comprising selecting short messages from at least one user of the plurality of users, the at least one user being a followcaster having at least a threshold number of subscribers.
4. The method of claim 3 , the identifying a segment in the media event further comprising:
identifying, using the at least one computing device, the segment in the media event using short message activity of at least one user identified as a followcaster having at least a threshold number of subscribers.
5. The method of claim 1 , further comprising:
selecting, using the at least one computing device, the sampling of short messages from a collection of short messages, the selecting comprising selecting short messages of a conversational-type.
6. The method of claim 5 , wherein the conversational-type message contains an indication that the message is directed to one or more users, the indication comprising an indicator linking the message sender and the one or more users.
7. The method of claim 1 , the identifying a segment in the media event further comprising:
identifying, using the at least one computing device, the segment in the media event using short messages identified as a conversational-type message.
8. The method of claim 1 , the identifying a segment in the media event further comprising:
determining, using the at least one computing device and the sampling of short messages, a plurality of term frequency scores for a term used in the sampling of short messages, each of the plurality of term frequency scores corresponding to a window of time of the media event and being indicative of a number of short messages containing the term in the corresponding window of time;
determining, using the at least one computing device and the sampling of short messages, a plurality of normalized frequency scores corresponding to the plurality of term frequency scores, for each term frequency score, the corresponding normalized frequency score comprising a ratio of the term frequency score and a corpus term frequency, the corpus term frequency indicative of a number of short messages in the sampling containing the term;
determining, using the at least one computing device and the plurality of normalized term frequency scores identified for the term, a maximum normalized term frequency score; and
identifying, using the at least one computing device, the segment from the window of time corresponding to the determined maximum normalized frequency score for the term.
9. The method of claim 1 , identifying at least one term taken from the sampling of short messages further comprising:
determining, using the at least one computing device and the sampling of short messages, a plurality of term frequency scores for a term used in the sampling of short messages, each of the plurality of term frequency scores corresponding to a window of time of the media event and being indicative of a number of short messages containing the term in the corresponding window of time;
determining, using the at least one computing device and the plurality of term frequency scores identified for the term, whether the term's frequency of use is higher relatively at a time corresponding to the identified segment; and
identifying, using the at least one computing device, the term as a term indicative of a context of the identified segment if the term's frequency of use is higher relatively at a time corresponding to the identified segment.
10. The method of claim 9 , wherein each of the term frequency scores comprises a normalized frequency score, the normalized frequency score comprising a ratio of a term frequency indicative of a number of short messages containing the term for the window of time and a corpus term frequency indicative of a number of short messages in the sampling containing the term.
11. A non-transitory computer-readable storage medium tangibly encoded with computer-executable instructions that when executed by a processor associated with a computing device perform a method comprising:
obtaining a sampling of short messages with content posted by a plurality of users, by sampling a collection of short messages with content posted during a broadcast of a media event's content;
identifying, using the sampling of short messages, a segment in the media event's content, identification of the segment comprising identifying an onset and an end of the identified segment in the media event's content using levels of short messaging activity detected using the sampling of short messages, the levels comprising a level used in identifying the onset and a different level used in identifying the end;
identifying at least one term taken from the content of the sampling of short messages, the at least one term indicative of a context of the identified segment in the media event content; and
annotating the media event's content using the at least one term identified from the content of the sampling of short messages.
12. The non-transitory computer-readable storage medium of claim 11 , annotating the media event's content further comprising:
annotating the segment identified in the media event's content using the at least one term identified from the content of the sampling of short messages.
13. The non-transitory computer-readable storage medium of claim 11 , further comprising:
selecting the sampling of short messages from a collection of short messages, the selecting comprising selecting short messages from at least one user of the plurality of users, the at least one user being a followcaster having at least a threshold number of subscribers.
14. The non-transitory computer-readable storage medium of claim 13 , identifying a segment in the media event further comprising:
identifying the segment in the media event using short message activity of at least one user identified as a followcaster having at least a threshold number of subscribers.
15. The non-transitory computer-readable storage medium of claim 11 , further comprising:
selecting the sampling of short messages from a collection of short messages, the selecting comprising selecting short messages of a conversational-type.
16. The non-transitory computer-readable storage medium of claim 11 , identifying a segment in the media event further comprising:
identifying the segment in the media event using short messages identified as a conversational-type message.
17. The non-transitory computer-readable storage medium of claim 11 , identifying a segment in the media event further comprising:
determining, using the sampling of short messages, a plurality of term frequency scores for a term used in the sampling of short messages, each of the plurality of term frequency scores corresponding to a window of time of the media event and being indicative of a number of short messages containing the term in the corresponding window of time;
determining, using the sampling of short messages, a plurality of normalized frequency scores corresponding to the plurality of term frequency scores, for each term frequency score, the corresponding normalized frequency score comprising a ratio of the term frequency score and a corpus term frequency, the corpus term frequency indicative of a number of short messages in the sampling containing the term;
determining, using the plurality of normalized term frequency scores identified for the term, a maximum normalized term frequency score; and
identifying the segment from the window of time corresponding to the determined maximum normalized frequency score for the term.
18. The non-transitory computer-readable storage medium of claim 11 , identifying at least one term taken from the sampling of short messages further comprising:
determining, using the sampling of short messages, a plurality of term frequency scores for a term used in the sampling of short messages, each of the plurality of term frequency scores corresponding to a window of time of the media event and being indicative of a number of short messages containing the term in the corresponding window of time;
determining, using the plurality of term frequency scores identified for the term, whether the term's frequency of use is higher relatively at a time corresponding to the identified segment; and
identifying the term as a term indicative of a context of the identified segment if the term's frequency of use is higher relatively at a time corresponding to the identified segment.
19. The non-transitory computer-readable storage medium of claim 18 , wherein each of the term frequency scores comprises a normalized frequency score, the normalized frequency score comprising a ratio of a term frequency indicative of a number of short messages containing the term for the window of time and a corpus term frequency indicative of a number of short messages in the sampling containing the term.
20. A computing device comprising:
a non-transitory storage medium for tangibly storing thereon program logic for execution by the processor, the program logic comprising:
obtaining logic executed by the processor for obtaining a sampling of short messages with content posted by a plurality of users, the sampling of short messages comprising a collection of short messages with content posted during a broadcast of a media event's content;
identifying logic executed by the processor for identifying, using the sampling of short messages, a segment in the media event's content, identification of the segment comprising identifying an onset and an end of the identified segment in the media event's content using levels of short messaging activity detected using the sampling of short messages, the levels comprising a level used in identifying the onset and a different level used in identifying the end;
identifying logic executed by the processor for identifying at least one term from the content of the sampling of short messages, the at least one term indicative of a context of the identified segment in the media event content; and
annotating logic executed by the processor for annotating the media event's content using the at least one term identified from the content of the sampling of short messages.