To answer my own question, I think this is the correct solution:
PREFIX
CONSTRUCT { ... }
WHERE {
BIND () {
SELECT ...
WHERE {
...
}
GROUP BY ...
}
}
Best,
Peter
> -----Ursprüngliche Nachricht-----
> Von: Wetz Peter [mailto:peter.w...@tuwien.ac.at]
> Gesendet: Dienstag, 24. November 2015 11:24
> An: users@jena.apache.org
> Betreff: BIND in CONSTRUCT query with sub-SELECT (SPARQL)
>
> Related to the answer of a previous question of mine (see [1]) I have a new
> question: I currently have a CONSTRUCT query with a nested SELECT which
> makes use of aggregates. What I now want to do is to construct new URIs in
> the CONSTRUCT statement.
>
> What I want to do is basically similar as can be seen in [2]. There I see that
> BIND is used in the WHERE class to construct new URIs which are then used in
> the CONSTRUCT clause. Unfortunately this approach does not work for my
> case (nested SELECT which resides in the CONSTRUCT's WHERE clause). The
> difference is, that in this example no nested SELECT is present.
>
> My current query looks like this (simplified):
>
> PREFIXES
> CONSTRUCT { ?s rdfs:label ?var . }
> WHERE {
> SELECT ?s (AVG(?single) as ?agg) ...
> WHERE {
> ...
> }
> GROUP BY ?s ...
> }
>
> The question is: Where to place a BIND statement which is used to bind
> values to variables, which are then used in the CONSTRUCT statement (e.g.
> ?var)?
>
> Thanks for your help.
>
> Best,
> Peter
>
> [1] https://mail-archives.apache.org/mod_mbox/jena-
> users/201511.mbox/%3C19440356.622107.1446464087017.JavaMail.yahoo@
> mail.yahoo.com%3E
> [2] http://mail-archives.apache.org/mod_mbox/jena-
> users/201111.mbox/%3c4ed66723.7030...@googlemail.com%3E
