Performance issue with textmatch

Support Bug
1

Hi,

i want to query stardog (6.1.2) on a bsbm ~35m dataset with a faceted-browsing like query, e.g.

PREFIX bsbm-inst: <http://www4.wiwiss.fu-berlin.de/bizer/bsbm/v01/instances/>
PREFIX bsbm: <http://www4.wiwiss.fu-berlin.de/bizer/bsbm/v01/vocabulary/>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>



SELECT  distinct ?o ?label ?desc {

    
    FILTER EXISTS {
        ?resource a bsbm:Product.
        ?resource bsbm:productFeature ?feature.
        ?resource bsbm:productPropertyTextual1 ?pf1.
        ?resource bsbm:productPropertyTextual2 ?pf2.
        ?review bsbm:reviewFor ?resource.

    }
    ?resource ?p ?o.
    
    FILTER (isiri(?o))
    optional {
        ?o rdfs:label ?label
    }

    optional {
        ?o rdfs:comment ?comment

    }

    ?o rdfs:label ?osearch.
    ?osearch <tag:stardog:api:property:textMatch> "da*".
    #FILTER (contains(?osearch,"da"))

}

limit 10000

performance dramatically drops, when using textmatch compared to contains.

  • contains ~1.1 s; 1600 results
  • textMatch ~400s; 437 results

As the results appear to be what is expected here, however the performance penalty is quite high.

The problem is much less pronounced, when the text-filtering is issued without other graph patterns, e.g.
PREFIX bsbm-inst: http://www4.wiwiss.fu-berlin.de/bizer/bsbm/v01/instances/
PREFIX bsbm: http://www4.wiwiss.fu-berlin.de/bizer/bsbm/v01/vocabulary/
PREFIX rdfs: http://www.w3.org/2000/01/rdf-schema#
PREFIX rdf: http://www.w3.org/1999/02/22-rdf-syntax-ns#

select distinct * {

    ?o rdfs:label ?oSearch.
    ?oSearch <tag:stardog:api:property:textMatch> "da*".
    #FILTER(contains(?oSearch,"da"))
}
limit 100000

Note:
The double use of rdfs:label might strike as odd, but as the query is generated, this might very well happen.

2

Hi Jörg,

Welcome to the forum. Thanks for the detailed report. Can you please share the query plans for the textMatch vs contains queries?

Jess