Cross media targeted message synchronization
Social media content items and references to events that occur therein are aligned with the time-based media events they describe. These mappings may be used as the basis for sending messages to populations of authors of content items, where the populations are determined based on whether the author has written a content item that refers to a specific TV show or advertisement. TV streams are monitored to detect when and where a specific advertisement for a particular advertiser is shown. Concurrently, social media streams are monitored for content items that refer to or are about specific TV shows and advertisements. Responsive to a specific advertisement being detected as being shown during a specific TV show, a message associated with the advertisement is sent to the authors of the content items associated with that TV show or advertisement. The messages can be transmitted while the advertisement is being shown.
1. A computer-executed method for sending messages to authors of social media content items, comprising:
identifying a plurality of candidate social media content items broadcasted by a social media source;
identifying a plurality of media events from a time-based media source, the time-based media source independent from the social media source, and the media events including a television media event and an advertisement media event;
for each of the candidate social media content items, determining, using a computer processor, a confidence score indicative of a probability that the candidate social media content item is relevant to the television media event;
aligning a subset of the candidate social media content items with the television media event based on the confidence scores;
forming a population of authors comprising authors of the subset of the candidate social media content items;
determining that the advertisement media event has aired during an airing of the television media event; and
responsive to the airing of the advertisement media event, sending the population of authors a message.
2. The computer-executed method of claim 1 , wherein determining that the advertisement media event has aired during the airing of the television media event comprises:
extracting event features from metadata annotations associated with the media events;
mapping the event features of the advertisement media event to the event features of the television media event, the event features including an airing time and date.
3. The computer-executed method of claim 1 , further comprising determining that the advertisement media event is currently airing.
4. The computer-executed method of claim 1 , wherein forming the population of authors comprises filtering the authors based on at least one selected from a group consisting of author demographic information, content of the candidate social media content items, and time when the candidate social media content items were created.
5. The method of claim 1 , further comprising:
extracting one or more features from the television media event, each feature describing an image or audio property of the television media event;
ranking a plurality of metadata instances based on a similarity between each metadata instance and at least one of the one or more features; and
annotating the television media event with a highest-ranked metadata instance to create an annotated event.
6. The computer-executed method of claim 5 , wherein determining the confidence score indicative of the probability that at least one of the candidate social media content items is relevant to the television media event further comprises:
extracting event features from the annotated event;
extracting social media features from the candidate social media content item; and
identifying a relationship between the event features and social media features, the confidence score of the candidate social media content item being based at least partially on the relationship.
7. The method of claim 6 , wherein identifying the relationship comprises identifying a correlation between a location specified by the event features and a location specified by the social media features and a correlation between an airing time specified by the event features and a publishing time specified by the social media features.
8. The method of claim 6 , wherein identifying the relationship comprises analyzing the event features and authority features included in the social media features to determine a probability that an author of the social media content item would generate content about an event having the event features.
9. The method of claim 5 , wherein each metadata instance includes content features extracted from one or more sample events associated with a different content domain.
10. The method of claim 9 , wherein ranking the plurality of metadata instance is further based on a similarity between the content domain associated with the metadata instance and a content domain associated with the television media event.
11. A system for sending messages to authors of social media content items, the system comprising:
a computer processor; and
a computer-readable storage medium storing computer program engines configured to execute on the computer processor, the computer program engines comprising:
an author store configured to identify a plurality of candidate social media content items broadcasted by a social media source;
an event airing engine configured to determine a plurality of media events from a time-based media source, the time-based media source independent from the social media source, and including a television media event and an advertisement media event;
a social media/event alignment engine configured to determine a confidence score for each of the candidate social media content items indicative of a probability that the candidate social media content item is relevant to the television media event and to align a subset of the candidate social media content items with the television media event based on the confidence scores; and
an audience population engine configured to form a population of authors comprising the authors of the subset of the candidate social content items;
a television show/ad overlap engine configured to determine that the advertisement media event has aired during an airing of the television media event; and
a message selection engine configured to send the population of authors a message responsive to the airing of the advertisement media event.
12. The system of claim 11 , wherein to determine that the advertisement media event has aired during the airing of the television media event, the television show/ad overlap engine is configured to:
extract event features from metadata annotations associated with the media events;
map the event features of the advertisement media event to the event features of the television media event, the event features including an airing time and date.
13. The system of claim 11 , wherein the television show/ad overlap engine is configured to determine that the advertisement media event has aired within the last minute.
14. The system of claim 11 , wherein the television show audience aggregation engine is configured to form the population based on at least one selected from a group consisting of author demographic information, content of the candidate social media content items, and time when the candidate social media content items were created.
15. The system of claim 11 , wherein the television show audience aggregation engine is configured to include in the population authors of candidate social media content items associated with the television media event, where the advertisement media event has aired during the television media event.
16. The system of claim 11 , further comprising:
a feature extraction engine configured to extract one or more features from the television media event, each feature describing an image or audio property of the television media event; and
a metadata alignment engine configured to:
rank a plurality of metadata instances based on a similarity between each metadata instance and at least one of the one or more features; and
annotate the television media event with a highest-ranked metadata instance to create an annotated event.
17. The system of claim 16 , wherein each metadata instance includes content features extracted from one or more sample events associated with a content domain.
18. The system of claim 17 , wherein ranking the plurality of metadata instances is further based on a similarity between the content domain associated with the metadata instance and a content domain associated with the television media event.
19. The system of claim 17 , wherein to determine the confidence score the social media/event alignment engine is configured to:
extract event features from the annotated event;
extract social media features from the candidate social content items; and
identify a relationship between the event features and social media features, the confidence score of the candidate social media content item being based at least partially on the relationship.
20. The system of claim 19 , wherein identifying the relationship comprises identifying co-occurring information within the event features and social media features.
21. The system of claim 19 , wherein identifying the relationship comprises:
identifying a correlation between a location specified by the event features and a location specified by the social media features, and
identifying a correlation between an airing time specified by the event features and a publishing time specified by the social media features.
22. The system of claim 19 , wherein identifying the relationship comprises analyzing the event features and authority features included in the social media features to determine a probability that an author of the social media content item would generate content about an event having the event features.
23. A non-transitory computer-readable storage medium storing instructions that, when executed by a computer processor, cause the computer processor to perform the steps of:
identifying a plurality of candidate social media content items broadcasted by a social media source;
identifying a plurality of media events from a time-based media source, the time-based media source independent from the social media source, and the media events including a television media event and an advertisement media event;
for each of the candidate social media content items, determining, using the computer processor, a confidence score indicative of a probability that the candidate social media content item is relevant to the television media event;
aligning a subset of the candidate social media content items with the television media event based on the confidence scores;
forming a population of authors comprising authors of the subset of the candidate social media content items;
determining that the advertisement media event has aired during an airing of the television media event; and
responsive to the airing of the advertisement media event, sending the population of authors a message.