OSLC Link Discovery Management - support reverse link lookup

[MartinUlrich]

Links are replicated to LDM in the format: subject, predicate, object. The LDM service shall use the predicate (e.g. satisfies) for filtering and in the triples returned.

An alternative approach could be, that LDM translates the predicate in to the reverse predicate (e.g. satisfied by), in order to consider the reverse predicate for filtering and in the triples returned.

Proposal is not to support reverse link lookup.

Rationale:

• LDM should not know anything about the semantics of the predicates, to keep the service simple
• semantics of predicates can easily be managed by the clients

[berezovskyi]

I think reverse link lookup is raison d'être for LDM. Let's assume there is a link somewhere in the system of the form:

<https://system_a/resource_a> oslc_rm:elaborates <https://system_b/resource_b> .

System B is expected to make a query to LDM of the following type:

?s oslc_rm:elaborates <https://system_b/resource_b> .

or even

?s ?p <https://system_b/resource_b> .

Is this the kind of lookup you wish to skip? Or are you proposing to skip the <https://system_b/resource_b> oslc_rm:elaboratedBy ?s inverse property search when a query for ?s oslc_rm:elaborates <https://system_b/resource_b> . is made? It was not immediately clear to me from your issue description.

[jamsden]

LDM should only do what the client asked for. That allows those clients to determine what they want to do with sameAs and inverses, both are things not supported by OSLC. From: MartinUlrich ***@***.***> Date: Tuesday, September 5, 2023 at 12:50 PM To: oslc-op/oslc-specs ***@***.***> Cc: Subscribed ***@***.***> Subject: [EXTERNAL] [oslc-op/oslc-specs] OSLC Link Discovery Management - support reverse link lookup (Issue #600) Links are replicated to LDM in the format: subject, predicate, object. The LDM service shall use the predicate (e. g. satisfies) for filtering and in the triples returned. An alternative approach could be, that LDM translates the predicate in ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. Report Suspicious <https://us-phishalarm-ewt.proofpoint.com/EWT/v1/PjiDSg!2q-qKtX1KRJyuc5TswsmJg8aP5xY6VrXOByk0jbFHKwG9ugtujuXcBa4V4EVZ5-JrtiuvhT8jQ2qm0hL6OwQe3k$> ‌ ZjQcmQRYFpfptBannerEnd Links are replicated to LDM in the format: subject, predicate, object. The LDM service shall use the predicate (e.g. satisfies) for filtering and in the triples returned. An alternative approach could be, that LDM translates the predicate in to the reverse predicate (e.g. satisfied by), in order to consider the reverse predicate for filtering and in the triples returned. Proposal is not to support reverse link lookup. Rationale: * LDM should not know anything about the semantics of the predicates, to keep the service simple * semantics of predicates can easily be managed by the clients — Reply to this email directly, view it on GitHub<#600>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAIQFKQ44P425VKOBOBJ5ILXY5J4PANCNFSM6AAAAAA4MCAF7Y>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[MartinUlrich]

?s oslc_rm:elaborates <https://system_b/resource_b> .
and
?s ?p <https://system_b/resource_b> .
shall be supported.

Searching for the inverse property shall be skipped.

[DavidJohnHoney]

I agree with Jim - it is up to clients to specify the link types and target URIs, and LDM should only do what the client asks.
We do need support for owl:sameAs as it's a common pattern - see https://jazz.net/wiki/bin/view/LinkedData/BestPractices