Hi Hubertus,
I would not expect that behavior, so probably the best thing to do is contact
ML support if you have not already done so. My understanding is that element
value queries use the same index data structures under the hood as word/term
queries, so I would not expect fewer results for a word query when compared to
a (nearly) equivalent element value query.
-Will
> On Feb 13, 2018, at 5:48 AM, Willuhn, Hubertus
> <hubertus.will...@dinsoftware.de> wrote:
>
> Hello Community,
>
> I have a question regarding a strange behavior difference between term
> queries and value queries.
>
> During some test for a new service, my team and I were trying to query
> documents based on a value query and the german language (xml:lang=”de”).
> However the query result was empty, but we do know there are documents that
> should match.
>
> So I test around a bit and found out that if you send a raw combined query
> (via the Java API) to the REST service of MarkLogic ,
> the server just ignores any language set via the query options.
>
> To make things clear for you, I created a simple test case that shows this
> behavior. In a fresh new database with default settings I created 4 test
> documents via the following XQuery:
>
> xdmp:document-insert("/demo1.xml", <root><element
> xml:lang="de">Zeitstand-Innendruckfestigkeit</element></root>);
> xdmp:document-insert("/demo2.xml", <root><element xml:lang="en">long-term
> hydrostatic strength</element></root>);
> xdmp:document-insert("/demo3.xml",
> <root><element>Zeitstand-Innendruckfestigkeit</element></root>);
> xdmp:document-insert("/demo4.xml", <root><element>long-term hydrostatic
> strength</element></root>);
>
> As you can see, the documents are tagged differently by language or have no
> tagging at all (which should result in English by default).
>
> Now I am searching for the document with the following command:
>
> search:resolve(<query xmlns="http://marklogic.com/appservices/search";
> xmlns:search="http://marklogic.com/appservices/search";>
> <and-query>
> <value-query type="string">
> <element ns="" name="element"/>
> <text>long-term hydrostatic strength</text>
> <weight>1.0</weight>
> </value-query>
> </and-query>
> </query>,
> <search:options xmlns:search="http://marklogic.com/appservices/search";>
> <search:page-length>300</search:page-length>
> <search:term>
> <search:term-option>lang=de</search:term-option>
> </search:term>
> </search:options>)
>
> So if you execute the search with the English word “long-term hydrostatic
> strength” and “<search:term-option>lang=en</search:term-option>” within the
> query options part,
> all is fine and it results in 2 found documents (for both value-query and
> term-query).
>
> <search:response snippet-format="snippet" total="2" start="1"
> page-length="300" xmlns:search="http://marklogic.com/appservices/search";>
> <search:result index="1" uri="/demo2.xml"
> path="fn:doc("/demo2.xml")" score="14336" confidence="0.5296452"
> fitness="0.7490314">
> <search:snippet>
> <search:match
> path="fn:doc("/demo2.xml")/root/element"><search:highlight>long-term
> hydrostatic strength</search:highlight></search:match>
> </search:snippet>
> </search:result>
> <search:result index="2" uri="/demo4.xml"
> path="fn:doc("/demo4.xml")" score="14336" confidence="0.5296452"
> fitness="0.7490314">
> <search:snippet>
> <search:match
> path="fn:doc("/demo4.xml")/root/element"><search:highlight>long-term
> hydrostatic strength</search:highlight></search:match>
> </search:snippet>
> </search:result>
> <search:metrics>
> <search:query-resolution-time>PT0.001277S</search:query-resolution-time>
>
> <search:snippet-resolution-time>PT0.001016S</search:snippet-resolution-time>
> <search:total-time>PT0.003034S</search:total-time>
> </search:metrics>
> </search:response>
>
> The same applies if you try to run this with German language and the German
> equivalent term “Zeitstand-Innendruckfestigkeit”. And the third scenario does
> also work: searching for
> the German term with no language constraint will result in one found document
> (with both term-query and value-query).
>
> However if you do above query for the English term with the language
> constraint “<search:term-option>lang=de</search:term-option>” within the
> query options,
> it results in no documents found for the term-query but in 2 found documents
> for the value-query!
> In other words it seems that the value-query just ignores the language
> constraint set by the query options.
>
> My question for this long story is: Is this a desired behavior or is this
> something I should report to MarkLogic support as a bug?
> Mit freundlichen Grüßen
>
> i.A. Hubertus Willuhn
> Informatiker
> Datenservice | XML-Technologie
>
> T +49 30 2601-2032| F +49 30 2601-42032
>
>
>
> Folgen Sie uns auf
> <image97879d.JPG>
>
> DIN Software GmbH, Am DIN-Platz, Burggrafenstraße 6, 10787 Berlin;
> http://www.dinsoftware.de; Registergericht: AG Berlin-Charlottenburg, HRB
> 28484; Geschäftsführer: Dr.-Ing. Mario Schacht
>
> Der Inhalt dieser E-Mail (einschließlich Anhängen) ist vertraulich. Falls Sie
> diese E-Mail versehentlich erhalten haben, löschen Sie sie bitte und
> informieren den Absender. Die DIN Software GmbH liefert Daten und erteilt
> Auskünfte nach Maßgabe einer Haftungsbeschränkung, die hier abrufbar ist. The
> contents of this e-mail (including attachments) are confidential. If you
> received this e-mail in error, please delete it and notify the sender. DIN
> Software GmbH provides data and information in accordance with its statement
> on the limitation of liability which is availablehere.
>
>
>
>
> _______________________________________________
> General mailing list
> General@developer.marklogic.com
> Manage your subscription at:
> https://urldefense.proofpoint.com/v2/url?u=http-3A__developer.marklogic.com_mailman_listinfo_general&d=DwICAg&c=IdrBOxAMwHPzAikPNzltHw&r=_thRNTuzvzYaEDwaA_AfnAe5hN2lWgi6qdluz6ApLYI&m=gIW0Gl9o2nkosOu8l8Bbfk9OiOGt7BS0S1xIIk_plzQ&s=TsnkIuTQ3VeBXzOZYMR8KQZqIH4tTJaorw9I9ZvRkXs&e=
_______________________________________________
General mailing list
General@developer.marklogic.com
Manage your subscription at:
http://developer.marklogic.com/mailman/listinfo/general
