Skip to content

Latest commit

 

History

History
89 lines (61 loc) · 4.23 KB

RECOMMENDATION.md

File metadata and controls

89 lines (61 loc) · 4.23 KB
Raw

Playfff - Recommendation

About

Playfff is the middleware to deliver recommendations for Play SRG mobile applications.

Compatibility

Since July 2018, Play Android (2.0.207 and more) and Play iOS (2.8.3-272 and more) applications currently use this recommendation service.

Media recommendation list for a media

The API doesn't not support paginations, therefore mobile applications didn't implement pagination. The media recommendation list must have at least 49 items, the 50th is the requested media, put on the first position of the list. If there is no recommended media, the returned list is empty.

Usages

By default, here are the used recommendation engines (RE). RTS_RECOMMENDATION_USED and SRF_RECOMMENDATION_USED environment variables can fallback to default Pfff RE recommendation engine.

  • continuousPlayback is used to display one proposition, with a count down and an autoplay.
  • relatedContent is used to display some propositions in a swimlane or a grid, without an autoplay.
BU Content Continuous playback Related content
RSI audios Pfff RE Pfff RE
RSI videos Pfff RE Pfff RE
RTR audios Pfff RE Pfff RE
RTR videos Pfff RE Pfff RE
RTS audios Pfff RE Pfff RE
RTS videos RTS RE RTS RE
SRF audios Pfff RE SRF RE
SRF videos Pfff RE SRF RE
SWI videos Pfff RE Pfff RE
Event videos N/A N/A
Swisstxt videos Pfff RE Pfff RE

Recommendation engine descriptions

RTS recommendation engine (RTS RE)

  • For RTS videos, it can ask rts-datalab.azure-api recommendation continuous_playback_endscreen service, without personalization.

SRF recommendation engine (SRF RE)

  • For SRF videos and SRF audios, it can ask IL-MediaList recommendation Recommended-byUrn service, without personalization.

Playfff On Demand recommendation engine (Pfff RE)

  • For RSI, RTR, RTS, SRF, SWI on demand videos and RSI, RTR, RTS, SRF on demand audios, it can ask Playfff recommendation.
  • Based on IL requests, without personalization. Here is how it works:
    • Get IL-Media. It returns an empty list if it's a LIVESTREAM or a SCHEDULED_LIVESTREAM.
    • Get IL-EpisodeComposition with last 100 episodes. Sort episodes with a published date ascending order, then episode ascending order if same published date. It's assumed that clips are ascendant sorting.
    • Determine if the media is a full length or a clip.
    • Separate in a full length list and a clip (audio only) list.
    • Get the requested media position in the related list. Split oldests and newests.
    • Recommendation list:
      • Newest medias in the date ascending order.
      • Then (ASCENDING_EPISODES_MAX value is set on the server, default is 25):
        • If more than ASCENDING_EPISODES_MAX episodes exists, oldest medias in date descending order.
        • Else (show has as many or fewer than ASCENDING_EPISODES_MAX episodes), oldest medias in the date ascending order.
  • It can get IL-MediaComposition if the media urn isn't found, and has not the CLIP type.
  • It does not return clips if VIDEO media type and standalone == false.
  • If clips are not in IL-EpisodeComposition, it fallbacks to full lengths.

Playfff Swisstxt video recommendation engine (Pfff RE)

  • For RSI, RTS, SRF swisstxt videos, it can ask Playfff recommendation.
  • Based on IL requests, without personalization. Here is how it works:
    • Get IL-MediaList from Liceventer for scheduled livestreams. Filter only current playable livestreams.
    • Get IL-MediaList from Liceventer for episodes. Filter only current playable replay content.
    • First livestream content, then replay content are returned.
  • The standalone == false parameter is ignored.

Scheduled livestream (event) URNs or other MAMs

  • No recommendation provided. It returns an empty list.

Media recommendation list for a user

The API doesn't not support paginations, therefore mobile applications didn't implement pagination. The media recommendation list must have at least 50 items.

RTS

  • With and without an userId, it asks rts-datalab.azure-api recommendation play_home_personal_rec service.

Other BUs

  • Nothing implemented.