In one aspect, an example method involves a computing device accessing first data representing a program schedule of a media program; accessing second data representing closed-captioning text of the media program; making a determination that first text of the program schedule has at least a threshold extent of similarity with second text of the closed-captioning text, where a portion of the media program was recorded proximate a time when a portion of the second data representing the second text of the closed-captioning text was generated; using the first text of the program schedule as a basis to select a portion of the program schedule; and responsive to at least the determination, storing third data indicative of a correlation between the selected portion of the program schedule and the recorded portion of the media program.