YouTube annotations and Video Indexing

Real purpose of YouTube annotations gets unnoticed as people dive into “birth of Long-Form video” discussions

There is quite a bit of noise in the blogspace about the recently-released YouTube Annotations feature and how it can be used to overcome the 10-minute video limitation on YouTube. You split the long episode into 10 minute parts and link them using the annotation at the end of each clip with the URL assigned to the annotation. Google Operating System blog was one of the first to comment on the news, CamcorderInfo has written about it and did again discuss the Long-Form trend, Search Engine Roundtable did a post and so did a few others.

What I feel has been missed is the real reason why Google added this feature to their popular video hosting service.

So far Computer Vision specialists from all around the world have been tackling the recognition, tagging and ranking problem for the videos. We know how to index and rank text-based information, but how do we do the same for videos, sounds, images?
At the moment factors that influence search rank of the videos are mostly based on the information outside the video, e.g. video tags, video title, video filename, text surrounding the video, text of the links to the video and user rating. Theoretically, a totally irrelevant video can have a high ranking in the search results if the information surrounding the video is relevant to the searched subject.

It makes sense to think that the Annotation feature added to YouTube is one of the intermediate steps taken by Google in order to understand the videos much better, provide higher quality ranking and reduce spamming with irrelevant content (questionable as knowing this the spammer could use annotations to fool the system). Unlike the video title, surrounding text and text of the links, the Annotations are part of the video and enable Google to better understand what exactly is being shown in the video.

Few people commented that annotations offered by YouTube lack any style and are not editable by viewers. Of course they could do a much prettier text overlay in any video editing program but that information won’t make it to the index and hence won’t make any contribution to the search ranking of this video.

So think about it twice before you go off to render the annotations in your favorite video editor! Besides, I am sure Google will add some sort of style editing and viewer annotations in the future releases. You will be able to go back and spice up your previous work at that stage.

Update, 1 Sept 2008: you should use the new captions feature to add subtitles.
The rest of the post still holds true – the right combination of captions, annotations, metadata and of course clear speech will help Google to index your videos much better.

Links on the Annotation feature:


Comments are closed.