How are murmurations nodes identified?

This is the forum for Murmurations.network

How are murmurations nodes identified?

wu-lee

I notice that there is no explicit ID field in any of the example schemas. I think there must be an implicit ID created for each one?

However, the things we want to index have their own unique IDs, which I have explicitly inserted a field for (this is a linked-data self-resolving URI).

There doesn't seem to be a way of marking a field as containing a unique value (either universally unique, or unique to the dataset). I would imagine this might be useful to do, do you have any thoughts about this?
Given one of our unique IDs, how would be then find the appropriate Murmurations node ID and link to it?

olisb

Hello Again - I hope I have understood your questions correctly - please let me know if any of the below does not make sense

wu-lee I notice that there is no explicit ID field in any of the example schemas. I think there must be an implicit ID created for each one?

Schemas are specifically designed to be entirely flexible, so there are no pre-defined mandatory fields at all.
If you need one, you can create an ID field and use it however you like.

However, the things we want to index have their own unique IDs, which I have explicitly inserted a field for (this is a linked-data self-resolving URI).

There doesn't seem to be a way of marking a field as containing a unique value (either universally unique, or unique to the dataset). I would imagine this might be useful to do, do you have any thoughts about this?

As above, create a new field via a PR in the Field Library (or simpler still, write the field directly into your schema) and then you can add "context" in the metadata to explain its use.

Given one of our unique IDs, how would be then find the appropriate Murmurations node ID and link to it?

After you search the index for nodes which validate against your chosen schema/s, and collect the profile URLs, you could search through the node profiles for an ID.

Does that help?
If not, please clarify the question/s

😉

wu-lee

To clarify the first bullet point in my question: let's say I've created a profile schema, and this includes an identifier field called my_id. Now, what stops two organisations creating profiles with the same value of my_id?

Nothing stops this, so far as I can tell. But I think this could be useful, perhaps essential in some cases.

olisb

wu-lee what stops two organisations creating profiles with the same value of my_id?

Nothing. Making a field contain only values which are unique throughout the entire use of the field would be very complex as far as I can tell...

wu-lee

To clarify the second bullet point in my question, I think that each registered profile has a page on the index server, and that page has an URL with what looks like a unique ID in it. In this example, the ID is cko085rpi00002a6cx3q3m0v2:

https://mpg.murmurations.network/api/p/cko085rpi00002a6cx3q3m0v2

This ID isn't part of the schema, it's assigned to the profile somehow by the index.

So if I have an organisation with a profile, and that's been indexed and an ID assigned, I might want to refer to the index server's page from somewhere else. How do I discover what the correct ID (or URL) is so I can reference it?

Also, how stable is that ID? For example, will it change if the profile content changes? Or if the profile location changes?

olisb

wu-lee I think that each registered profile has a page on the index server,

Not exactly. Each node profile has a unique URL (which is registered in the Index) - there are no 'pages' in the index, just links to node profiles

and that page has an URL with what looks like a unique ID in it. In this example, the ID is cko085rpi00002a6cx3q3m0v2:
https://mpg.murmurations.network/api/p/cko085rpi00002a6cx3q3m0v2

The ID you list is simply the end portion of the unique URL which is registered in the index, for this specific profile.
In the example you use the node profile URL was generated by the Murmurations Profile Generator, because I chose to host that specific profile with the MPG

Other profiles will not have that kind of ID, because they are hosted elsewhere. e.g. https://www.open.coop/open.json

This ID isn't part of the schema, it's assigned to the profile somehow by the index.

No, it's assigned to the node profile, by the MPG, when the user chooses to host their profile via the MPG

So if I have an organisation with a profile, and that's been indexed and an ID assigned, I might want to refer to the index server's page from somewhere else. How do I discover what the correct ID (or URL) is so I can reference it?

The question doesn't really make sense, because it's not how the Index works.
To find a list of profiles which match a certain schema you can search the index (for node profiles which validate against the chosen schema/s), collect the list of their unique URLs, then visit those URLs and download the data.

In Murmurations the node profile URL is the unique ID.

Also, how stable is that ID? For example, will it change if the profile content changes? Or if the profile location changes?

URLs which are generated by MPG do not change when the node profile is updated.

wu-lee

Ok, I think I see, there are two cases here:

Profiles hosted on 3rd party servers like https://www.open.coop/open.json
Profiles hosted on the profile generator server mpg.murmurations.network

The latter is just a convenience for people without a place to host their profile, and there isn't an URL like this created for the former case. In both cases, the profile ID is the profile URL, and if you know that you can also use that to reference it.

Presumably a consequence of that is: if your profile's URL changes at all, it must be re-registered with the index server, and will not be identified as the same thing any more, unless perhaps there's some other way to infer that (like a schema-specific ID field).

olisb

wu-lee Ok, I think I see, there are two cases here:
Profiles hosted on 3rd party servers like https://www.open.coop/open.json
Profiles hosted on the profile generator server mpg.murmurations.network

The latter is just a convenience for people without a place to host their profile, and there isn't an URL like this created for the former case. In both cases, the profile ID is the profile URL, and if you know that you can also use that to reference it.

Yes - exactly.

wu-lee Presumably a consequence of that is: if your profile's URL changes at all, it must be re-registered with the index server, and will not be identified as the same thing any more,

Correct - it would be good if you deleted the old profile from the Index, but clearly there is no way to enforce that.

unless perhaps there's some other way to infer that (like a schema-specific ID field).

If it was exactly the same profile, referencing the same schema/s, aggregators could probably match them them, and even infer greater 'authority' if the URL field in the profile matched the URL where the file was hosted. This is one of the mechanisms we hope to employ to limit spam and fake submissions, as mentioned in the white paper.

geoffturk

olisb If it was exactly the same profile, referencing the same schema/s, aggregators could probably match them them, and even infer greater 'authority' if the URL field in the profile matched the URL where the file was hosted.

The URL that infers 'authority' is what we are calling the primary_url in V2 of the protocol.

https://github.com/MurmurationsNetwork/MurmurationsLibrary/blob/06f37656fdd6861f6509ccfe64a78106bc6cb69d/fields/primary_url.json

If the profile_url (https://www.open.coop/open.json) is located at the primary_url (https://www.open.coop), it should be viewed as the definitive source about information related to the org operating the primary_url.

This is the forum for Murmurations.network

A free forum powered by FreeFlarum (remove this footer)
Terms & Conditions | Privacy Policy