NER to include endpoint of resource

[pheyvaer]

When you perform NER on a text, you get the resources of the entities. However, when more information is needed about a resource, by using e-Link, the endpoint is needed, because we need the endpoint to create a template.

Is it possible to add the endpoint for each resource to the NIF? This is especially important when the NIF is being used by multiple services, as a certain service might not know on which dataset e-entity was performed earlier on in the process.

[m1ci]

Is it possible to add the endpoint for each resource to the NIF?

By adding this info to each entity will increase the NIF size significantly.

What if we better documented the templates. So that we include information such as: "this template can be used if you performed NER using the DBpedia dataset?". Will this help?

[pheyvaer]

But then you still have to know which dataset was used for e-entity, and this is not always the case.

[m1ci]

well, maybe you can derive the dataset by looking at the entity URIs? http://dbpedia.org/resource/ for DBpedia http://orcid.org/ for ORCID ...

[pheyvaer]

This is not possible, as the entity URI do not have to be related to the endpoint URI.

[m1ci]

true, but also entity URI does not need to be present in any endpoint.

Also note that for entity linking is used data which not necessarily should be deployed in some endpoint. So there are two assumptions here: 1) the client should be aware about the entity links from which domain they come 2) the client should be aware in which datasets these entities are described

Again, for 1) we can not provide endpoint URI since, as I wrote early in the comment, not necessarily the entity URI is present in an endpoint. This is up to the client. As for 2), the client should check the list of available templates and the datasets behind them.

[m1ci]

IMO this is big and open research question: know in which endpoint a resource can be found. Closing the issue.