Should we summarise results as time or mentions?

[dcabo]

When searching for a term, like "Rajoy", we initially thought we'd summarise the results as "Rajoy was talked about during 32 minutes". But, if we're not able to segment the full transcription is meaningful blocks (#7), what we have left are sentences. We currently add the duration of each sentence, taken from the start/end times of the captions, but I'm not sure this is too meaningful, because not every sentence in a given block is going to mention "Rajoy" again and again.

When Vox analised the media coverage of Democratic candidates it didn't overcomplicate it: it used as unit the 15-second clips that the Internet Archive offers, and instead of talking of "minutes and seconds" it talks about "mentions".