US 11,057,457 B2
Television key phrase detection
Andrew J. Carroll, San Francisco, CA (US); Richard Douglas Whitcomb, Jr., San Francisco, CA (US); Wangfan Fu, San Francisco, CA (US); and Thanabordi Jianrungsang, San Francisco, CA (US)
Assigned to Twitter, Inc., San Francisco, CA (US)
Filed by Twitter, Inc., San Francisco, CA (US)
Filed on May 23, 2019, as Appl. No. 16/420,944.
Application 16/420,944 is a continuation of application No. 15/853,384, filed on Dec. 22, 2017, granted, now 10,326,829.
Application 15/853,384 is a continuation of application No. 14/187,056, filed on Feb. 21, 2014, granted, now 9,876,848, issued on Jan. 23, 2018.
Prior Publication US 2019/0281109 A1, Sep. 12, 2019
Int. Cl. H04L 12/70 (2013.01); H04L 29/08 (2006.01); G06K 9/00 (2006.01)
CPC H04L 67/10 (2013.01) [G06K 9/00536 (2013.01)] 24 Claims
OG exemplary drawing
 
1. A method comprising:
accessing a media event comprising an image that includes an unidentified phrase, wherein the media event is associated with a particular subject matter domain;
processing the unidentified phrase of the image using a character recognition algorithm to generate a set of candidate phrases;
determining a set of social media content items associated with the particular subject matter domain of the media event, wherein the determining includes identifying, in a collection of social media content items, the set of social media content items that each have content that matches a particular domain ontology indicating a respective vocabulary specific to the particular subject matter domain;
determining, for each candidate phrase in the set of candidate phrases, a respective phrase score based on the number of appearances of the candidate phrase in the set of social media content items;
selecting a candidate phrase associated with a highest phrase score as a final phrase representing the unidentified phrase; and
storing, in a storage medium, an association between the final phrase and the media event.