SPARQL Query for distance between nodes

[Marie Valadez]

I have searched and tested out multiple ways to get the distance between two nodes. I want a query that will show the child concepts and how far away they are from the current concept so that I can create a table on a form showcasing this.

The following query works if a ?sub concept does not have two connection through skos:broader. But once a ?sub has two parent concepts with the skos:broader it overcounts. I have included a basic example diagram on the issue I am dealing with. Any ideas on how to get an accurate count?

SELECT ?sub  (count(distinct ?mid) as ?distance)
WHERE {
  $this ^skos:broader* ?mid .
  ?mid ^skos:broader+ ?sub .
}
group by $this ?sub

[Holger Knublauch]

Hi Marie,

we do have some built-in functions including spif:shortestObjectsPath and swa:shortestPathsBetweenNodes that may help.

To clarify your requirements, is it true that one of the nodes is always an (indirect) parent of the other node, or does the algorithm also need to walk in one direction and then in another direction of the tree structure?

And are we talking about tree structures at all, or arbitrary graphs?

Holger

--
The topics of this mailing list include TopBraid EDG and related technologies such as SHACL.
To post to this group, send email to topbrai...@googlegroups.com
---
You received this message because you are subscribed to the Google Groups "TopBraid Suite Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to topbraid-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/be2ebc41-1fa7-4a70-a3ca-d8ee2cf575ddn%40googlegroups.com.
<stepcount-Page-2.drawio.png>

[Marie Valadez]

The way skos:broader is being used in our case is a tree like structure with some having multiple parents that can be broader but will point back to the same main broad concept which branches out to its children, grandchildren, etc. I need the shortest path from whichever instance is selected to all of its descendants. All will have the connection to each other through the skos:broader within the family tree (some can have more than 1 direct skos:broader concept (parent)). 

Do you have any example documentation on how those functions can be utilized? 

You received this message because you are subscribed to a topic in the Google Groups "TopBraid Suite Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/topbraid-users/CUgBING_53U/unsubscribe.
To unsubscribe from this group and all its topics, send an email to topbraid-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/D618E05A-6068-428F-B8DA-77C9D6C52A96%40topquadrant.com.

[Holger Knublauch]

Here is an example:

where $this is some skos:Concept deep in the hierarchy such as g:Vienna.

For easier copy and paste:

PREFIX g: <http://topquadrant.com/ns/examples/geography#>

PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

PREFIX spif: <http://spinrdf.org/spif#>

SELECT *

WHERE {

BIND (spif:shortestObjectsPath(g:Vienna, skos:broader, g:Europe) AS ?path) .

}

HTH

Holger

To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/CAH4XeR92ty0CU3k55_fYc%3DA_Vmjxy64b8vTcDKDU3uBUx9KjQA%40mail.gmail.com.

[Marie Valadez]

Thank you for the example use case. What does g:Europe represent in this case? Is that the ending node? Is it necessary to specify the ending node? For instance if I wanted to find all the descendants of a concept using the inverse of skos:broader and the distance each one is away from the current concept, then I wouldn't want to specify a single child concept. 

To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/F370933F-CA50-4DFA-8FC6-622FF0020E86%40topquadrant.com.

[Holger Knublauch]

Yes, g:Europe is the end node and is optional. If no end node is specified, it will stop at whenever a node doesn't have further values of skos:broader.

You can use spif:shortestSubjectsPath for the inverse direction.

To find the paths to each child, loop over the children:

SELECT *

WHERE {

    ?child skos:broader g:Europe .

    BIND (spif:shortestObjectsPath(?child, skos:broader, g:Europe) AS ?path) .

}

For anything more complex, ADS JavaScript is often a good choice as it gives you more control:

https://datashapes.org/active/

Holger

On 19 Feb 2024, at 12:35 am, Marie Valadez <meval...@gmail.com> wrote:

Thank you for the example use case. What does g:Europe represent in this case? Is that the ending node? Is it necessary to specify the ending node? For instance if I wanted to find all the descendants of a concept using the inverse of skos:broader and the distance each one is away from the current concept, then I wouldn't want to specify a single child concept. 

On Sun, Feb 18, 2024 at 3:24 AM Holger Knublauch <hol...@topquadrant.com> wrote:

Here is an example:

To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/CAH4XeR-6uzQ0gdaRe4q5Fry8%2BvLtokh3%3D1MbmcLnYsQbxRz0jQ%40mail.gmail.com.

[Marie Valadez]

Thank Holger. 

Is there a way to return a count with this that shows the distance?

[Holger Knublauch]

On 20 Feb 2024, at 6:11 pm, Marie Valadez <meval...@gmail.com> wrote:

Thank Holger. 

Is there a way to return a count with this that shows the distance?

Sure. The function returns a ?path string with one space between path segments.

You can use

BIND (REPLACE(?path, "[^ ]", "") AS ?spaces) .

BIND (STRLEN(?spaces) AS ?depth) .

to count the spaces.

Holger

To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/bb6a78ff-0569-49a0-9e3f-da3e0e97fb07n%40googlegroups.com.

[Marie Valadez]

Thank you so much! Appreciate all your help. 

[steveray...@gmail.com]

Holger, this thread was very interesting for me to read, because it addresses a navigational problem we're looking at. However, some of my colleagues do not use the TopQuadrant tools, so they don't have spif:shortestObjectsPath available. Am I correct that this was coded in another language? Would the best approach be to tackle this using the Javascript extension to SHACL mechanism? (Or even better, do you have the Javascript code?).

[Holger Knublauch]

Hi Steve,

this is implemented in Java, but yes there is no rocket science and it could be reimplemented in other languages. Basically do a breadth-first traversal.

Holger

To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/e84cc166-8a54-4b20-b460-668db07dc178n%40googlegroups.com.

[steveray...@gmail.com]

Thanks. Would you also agree that this approach is better than trying to do something in native SPARQL? I have always seen SPARQL as better suited to declarative, set-based problems rather than navigational ones.

[Matt Goldberg]

On a semi-related note, tasks like this are why I suggested a pathfinding extension to the SPARQL specification such that these types of operations would be supported across implementations https://github.com/w3c/sparql-dev/issues/191

On Thursday, March 7, 2024 at 11:52:52 AM UTC-5 Holger Knublauch wrote:

[Holger Knublauch]

If SPARQL could express this then the extra functions wouldn't be needed. So I guess this goes beyond what SPARQL can do.

I guess Matt just sent something along those lines.

Holger

To view this discussion on the web visit https://groups.google.com/d/msgid/topbraid-users/a4d306a2-2c00-45cc-ae45-50733fc80a95n%40googlegroups.com.