Podcastindex-org / podcast-namespace

A wholistic rss namespace for podcasting
Creative Commons Zero v1.0 Universal
389 stars 116 forks source link

Proposal: metadata tags <podcast:book>, <podcast:game>, <podcast:tv_show> #180

Open PofMagicfingers opened 3 years ago

PofMagicfingers commented 3 years ago

This is a transfered proposal from our former project podCloud/podcast-ext. See #173 for details.

Episode metadata tags

Any episodes can contains an infinite amount of tags to provide more data about the content of the episode.

This could include podcast:person from #96 if the episode is talking about a person (and said person is not a guest of the episode).

any object, a car, a computer, a furniture, a puree press etc..

This duplicates #138

Although we could use <podcast:link> from #176

With some childrens : podcast:location, podcast:website|link

Can use some childrens used at channel level : podcast:id, podcast:social etc. ( I really like the name of this one :grin: )

Some refinements are needed :

IMHO, this would be a big added value to rss feeds. It could allow podcasters to precisely tag what they where talking about in their episode, and allow podcast players to search by topic, like YouTube does on some content :

image

image

Bigaston commented 3 years ago

For the ID of game, maybe we can add the IGDB id or slug? I think it can work very well

PofMagicfingers commented 3 years ago

Yes! I guess we could use thetvdb for TV shows and imdb for movies (or any other more "opensourcy" alternative)

Le mar. 2 févr. 2021 à 18:27, Bigaston notifications@github.com a écrit :

For the ID of game, maybe we can add the IGDB id or slug? I think it can work very well

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Podcastindex-org/podcast-namespace/issues/180#issuecomment-771816735, or unsubscribe https://github.com/notifications/unsubscribe-auth/AADST7JALPDS3AQRMDFLYATS5AYYVANCNFSM4W7BLCEA .

swschilke commented 3 years ago

I would not use multiple tags that kind of express the same, a reference of or to something. Couldn't it be a "reference" tag with a value's like book, movie, show, game etc., a description, and then a referring URL/Reference.

PofMagicfingers commented 3 years ago

I would not use multiple tags

Yeah that was something I had in mind, maybe we could merge them into something like : <podcast:topic type="[tv|movie|music|etc]" id="[id that make sense given type]" url="[url] " [some other attr] />

ID could be in the form of : isbn:134432 or tvdb:242 and such.

We should discuss what attributes can work for all types. artist or author could work for games, TV shows, music, and movies, which are IMHO the most important/useful types.

We could add a date attributes that works for release dates, events and such.

Maybe we could move the title/name as text content but that prevent us to add children elements to be more specific or extend metadata.

theDanielJLewis commented 3 years ago

Whatever tags come from this should probably be in the episode metadata JSON with a master reference from the feed. Like <podcast:metadata url="…" />.

PofMagicfingers commented 3 years ago

In that case, maybe we should go for something like JSON-LD and Schema.org for metadatas. We could do an array of JSON-LD objects.

It does have many types : https://schema.org/Book https://schema.org/Movie https://schema.org/MusicRecording https://schema.org/TVSeries

It supports extensions, and is already supported on webpages by Google and such

Bigaston commented 3 years ago

I for and against the use of JSON metadata. For me it can be good to have it on RSS and in the same time in the JSON. Like that parser can just parse the RSS and don't have to do another request

PofMagicfingers commented 3 years ago

Using an external file for metadatas could be a good solution to limit bandwith usage. You get the main data in the RSS feed, and when you download or open an episode, the client download the metadata (topics, chapters, etc), only when needed.

swschilke commented 3 years ago

I just mentioned that I think it's a bad idea to split meta data into several files is error-prone.