Re: EXISTS variants

the meeting on 11.7 discussed whether the scope of the bindings implied by exist solution variables should be "shallow" or "deep".
the consensus was that the distinction would affect just expressions which include a select, as its projection can affect variable scope.

the question arose, how much can that matter.
from the queries performed on wikidata, practically, very very little.

of the queries archived by wikidata, just five included a select for in the exists form.
all were the initial operator in the form.

the attached file enumerates the variants.

best regards, from berlin,

> On 10. Jul 2025, at 19:12, James Anderson <anderson.james.1955@gmail.com> wrote:
> 
> the attached csv describes the nature of the 566 exists variants present in the wikidata sparql queries.
> the "operators" column exclude the exists operator itself, counts one for each bgp, one for each for each statement pattern and one join for each pair of statement patterns.
> 
> 
> 
> On Thu, 10 Jul 2025 at 18:37, James Anderson <anderson.james.1955@gmail.com> wrote:
> as a further excursion in this direction, i applied a revised version of the extraction process to the archived wikidata queries.
> these, as described by the ICCL Wikidata query archive
> 
>     https://iccl.inf.tu-dresden.de/web/Wikidata_SPARQL_Logs/en
> 
> there were a total of 1,310,678,852 SPARQL queries, of which 626,080 included exists operators.
> the attached graph depicts the operator count distribution
> 
> 
> 
> On Tue, 24 Jun 2025 at 10:02, James Anderson <anderson.james.1955@gmail.com> wrote:
> good morning;
> 
> one of the topics last week was the possible variations evident in how authors use the exists operator.
> 
> in order to provide some information, i looked at the queries recorded for several sites to which i have access.
> the variants and the nodejs script to extract the data are present in a fork of the SPARQL-exists github repository.
> 
>     https://github.com/datagraph/SPARQL-exists/tree/main/test-tools.
> 
> the top directory contains the .mjs script.
> the query-statistics directory includes the results from individual hosts and an aggregate csv file.
> 
> the queries have been anonymized to hide restricted information and to make the resemblances stand out.
> the analysis is context-free.
> 
> best regards, from berlin,
> ---
> james anderson | james@dydra.com | https://dydra.com
> 
> 
> <query-statistics.csv>

---
james anderson | james@dydra.com | https://dydra.com