CASopedia project requirements

[zzacharo]

Context

CASopedia project (CERN Accelerator schools) is an upcoming website aiming to promote the lectures recorded from the CERN Accelerator schools around the world.

Currently, they have a collection on CDS that we need to ensure that is part of our collected data to migrate.

We had a meeting with them (indico event) and below are the requirements gathered.

Requirements

Captions search on video (search only from the video page)

• It would be nice to be able to show the transcription and users to follow it/jump on the specific point in the video (as per youtube)
  • See also prototype.
• Take into account the language.
• Users should be able to search inside the transcription and once they find a relative term they could click and jump on the video point in time
• See example on current CDS.

Point in time jumps

We need to ensure that a video in the landing page can start from a specific point in time. This exist when embedding a video.

• We need to be able users to share links of the format: https://videos.cern.ch/record/12345?t=123

Display and search by slides on the player

We need to display the slides of a specific lecture in a way that people can click on them and jump on the specific point in time on the video.

• We should design a way for users to be able to upload slides/images and a way to link each image with a start and end time. For example:
  • Support a new .vtt file and associate each section with a file
  • Make the extraction of frames toggleable and allow users to upload their images. The challenging part is to find a way for people to associate each file with the start,end time
  • Implement the slides as part of a new feature to support chapters (See also prototype ). Take inspiration from how youtube is showing the chapters.

Ideas

Maybe we can introduce chapters

CAS access control

• Evaluate if a separate category needs to be created or just align on setting up explicitly the necessary egroups via the publishing workflow that the recording service team applies.

[zzacharo]

For captions search, see also the implementation from the recording service team for inspiration: https://github.com/polimediaupv/paella-basic-plugins/blob/main/src/plugins/es.upv.paella.findCaptionsPlugin.js

[gasparuben]

Thumbs generation on Opencast looks good for a 60x40 video: https://lecturemediaqa.cern.ch/2006/429456/.

[zzacharo]

Draft prototype branch: https://github.com/zzacharo/cds-videos/tree/thumbnails-jump-on-video

[zubeydecivelek]

Transcriptions search PR:
#2081

[zzacharo]

Experimented a bit with Claude AI and here is a #2086 with various changes and a summary. It is deployed on dev environment if you want to test it out.

@zubeydecivelek maybe check the code and see if we can merge it. It includes also a solution about the slides by adding chapters. You can take inspiration from the changes if you see that are complicated and create a separate PR.

Regarding chapters:

• User adds in the description of the deposit the timeline a specfic format. Example:

00:00 Slide 1
01:00 Slide 2
02:00 Slide 3
...

• On publish we call a new task to extract the chapters and ensure that we take a snapshot of the video at each chapter start time
• The chapters are displayed in the details page and the user can also toggle them in the same box as the transcriptions.
• On each edit of the description if there is a change on the timelines, we should rerun the chapter extract task when the user publishes.