Signposting and Content Negotiation

[Tobias Schweizer]

Hi there,

I have first heard of signposting when reading through 'FAIR Assessment Tools: Towards an "Apples to Apples" Comparisons' (https://zenodo.org/doi/10.5281/zenodo.7463420).

My initial understanding was that it referred to the HTML header (HTML is a terrible format for machine agents) but now I realise that signposting can also be implemented in the HTTP response header (https://signposting.org/FAIR/#typedlinks).

I have tried some of your examples (https://signposting.org/adopters) and I am thrilled (some of them are broken though) as this is exactly what I was looking for.

I initially started to search for examples for reactive or agent-driven content negotiation but actually found none. In this form of content negotiation, the server should send the client a list of available representations the client can choose from. And now I saw signposting in the HTTP header as a list of links.

So could these approaches be combined? I think there might be differences:

 - Content negotiation assumes that there are different representations of the same resource whereas signposting indicates the semantic difference (citeas, item etc.)

- Content negotiation uses the same URL with different accept headers sent by the client whereas signposting uses different URLs

- Could signposting happen at the DOI level like content negotiation (and not the actual repository link)?

I have a use case where I resolve DOIs using content negotiation to obtain RDF metadata about the resource because I am interested in author PIDs like ORCIDs. The obstacle is the fact that different DOI registration agencies support different content types (https://citation.crosscite.org/docs.html#sec-4). So depending on the registration agency, some previous knowledge ist required. 

It would be great if I could make a request for a given DOI and get back an HTTP response header telling me about the different available representations, including author ORCIDs. This way, I would not have to make a second request.

Let me know if I should provide some more details. Thanks for your feedback.

Tobias

[Herbert Van de Sompel]

On Jan 8, 2024, at 08:11, Tobias Schweizer <tobias.s...@switch.ch> wrote:

Hi there,

I have first heard of signposting when reading through 'FAIR Assessment Tools: Towards an "Apples to Apples" Comparisons' (https://zenodo.org/doi/10.5281/zenodo.7463420).

My initial understanding was that it referred to the HTML header (HTML is a terrible format for machine agents) but now I realise that signposting can also be implemented in the HTTP response header (https://signposting.org/FAIR/#typedlinks).

I have tried some of your examples (https://signposting.org/adopters) and I am thrilled (some of them are broken though)

I will check and fix. Thanks for noticing. 

as this is exactly what I was looking for.

I initially started to search for examples for reactive or agent-driven content negotiation but actually found none. In this form of content negotiation, the server should send the client a list of available representations the client can choose from. And now I saw signposting in the HTTP header as a list of links.

So could these approaches be combined? I think there might be differences:

 - Content negotiation assumes that there are different representations of the same resource whereas signposting indicates the semantic difference (citeas, item etc.)

If various representations are available for a same resource, then this can also be handled using Singposting. For example, if a content resource (linked from landing page via “item” link type) is available in both eg HTML and pdf then two “item” links can be provided with the same target URI but different values for the “type” attribute. That attribute conveys the media type of the target resource, and, in this example would convey text/html and application/pdf, respectively. That way, a client is informed how it can negotiate to obtain its preferred representation.

- Content negotiation uses the same URL with different accept headers sent by the client whereas signposting uses different URLs

See above

- Could signposting happen at the DOI level like content negotiation (and not the actual repository link)?

Signposting is about the custodian of research objects providing the links because that custodian is in the best position to provide authoritative information pertaining to its own objects. 

I have a use case where I resolve DOIs using content negotiation to obtain RDF metadata about the resource because I am interested in author PIDs like ORCIDs. The obstacle is the fact that different DOI registration agencies support different content types (https://citation.crosscite.org/docs.html#sec-4). So depending on the registration agency, some previous knowledge ist required. 

It would be great if I could make a request for a given DOI and get back an HTTP response header telling me about the different available representations, including author ORCIDs. This way, I would not have to make a second request.

I am in no position to speak on behalf of DOI registration agencies. Signposting aims at achieving what you are after by providing links in a decentralized manner, at the end of the repositories that manage research objects. 

Thanks for your interest in Signposting.

Greetings

Herbert 

Let me know if I should provide some more details. Thanks for your feedback.

Tobias

--
You received this message because you are subscribed to the Google Groups "signposting" group.
To unsubscribe from this group and stop receiving emails from it, send an email to signposting...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/signposting/7ca8a1f3-e536-4e6d-8e72-64cf8d31c6f6n%40googlegroups.com.