1

I'm looking for a way to identify (as many as possible) YouTube videos on a certain topics based on a keywords occurring in their titles.

YouTube Data API provides a way to search for videos based on the keywords but limits results up to 500 titles only.

Common Crawl indexes YouTube.com domain but it contains information only about page URLs and not its HTML <title> while YouTube.com videos do not have their titles in URL (basicaly it's https://www.youtube.com/watch?v=v01QZUxYg62).

There's YouTube 8M dataset of segmented videos but it seems an overkill to just get the titles. Is there any other corpora of up-to-date YouTube video titles?

1

You can use youtube-dl to make queries, and then dump just the metadata as a json, without downloading the video. Then you can parse the (huge) json to get just the ids and titles and categories, or any other field with jq

youtube-dl --no-warnings --dump-single-json "ytsearch1:mike birbiglia it's so easy to be on time" | jq -r '.entries[] | "\(.id)\t\(.title)\t\(.categories)"'

returns tab-separated id, title, categories

W-_M4r-gKI0 Mike Birbiglia - It's So Easy To Be On Time ["Comedy"]

my sources: metadata + jq and ytsearch

fyi, the link you shared wasn't available for me

btw, it's a funny video

2
  • 1
    Thanks. To get more results, e.g., 100, use ytsearch10:Obama
    – dzieciou
    Mar 2 at 7:08
  • And the link to YouTube video I've provided is fake.
    – dzieciou
    Mar 2 at 16:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.