Duplicate results in alphabetical listing

[tfrancart]

(using code from master branch)

I test SKOSMOS with a multilingual vocabulary, specifically the Unesco thesaurus available in SKOS at http://skos.um.es/unescothes/. I use Jena Fuseki (1.3.0) with JenaText dialect.
Concepts appear duplicated in the alphabetical listing on the left, as shown below :

Looking at the underlying SPARQL query, I see it looks like this :

PREFIX owl: <http://www.w3.org/2002/07/owl#>
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
PREFIX text: <http://jena.apache.org/text#>
SELECT ?s ?label ?alabel WHERE {   
    GRAPH <http://graph> {     
        {       
            { 
                ?s text:query (skos:prefLabel 'C*' 100000) 
            }
            ?s skos:prefLabel ?label .
            FILTER ( 
                strstarts(lcase(str(?label)), 'c')
                &&
                langMatches(lang(?label), 'en')
            )
        }
        UNION
        {
            { 
                ?s text:query (skos:altLabel 'C*' 100000) 
            }
            {
                ?s skos:altLabel ?alabel .
                FILTER (
                    strstarts(lcase(str(?alabel)), 'c')
                    &&
                    langMatches(lang(?alabel), 'en')
                )
            }
            {
                ?s skos:prefLabel ?label .
                FILTER (
                    langMatches(lang(?label), 'en')
                )
            }
        }
        ?s a ?type .
        FILTER NOT EXISTS {
            ?s owl:deprecated true
        }
    }
    VALUES (?type) { (<http://www.w3.org/2004/02/skos/core#Concept>) }
}
ORDER BY LCASE(IF(BOUND(?alabel), STR(?alabel), STR(?label)))

It returns duplicate rows, and probably should have a DISCTINCT keyword added.

[tfrancart]

The problem is really tied to the Jena text search, because if using SPARQL generic queries, the alphabetical listing is displayed correctly without duplicates.

[osma]

Thank you for reporting this and also testing without jena-text.

My guess is that this is due to changes introduced in the jena-text module of Jena 3.0.0 / Fuseki 1.3.0 / Fuseki 2.3.0. The text index can now optionally report more information (scores and original literals), but some low-level suppression of duplicates had to be skipped due to the richer information - I happen to know because I made those changes in jena-text.

We are currently still running Fuseki 1.1.x so hadn't yet noticed the problem. Will have to test with newer Fuseki. As you say, fixing this might be as easy as adding a DISTINCT clause to the query as a simple fix. You can of course try using a slightly older Fuseki as well, at least it's better than not using a text index at all.

To actually make use of the new features in jena-text 3.0.0 is described in #273, but it will not yet be implemented in the Skosmos 1.2 development cycle.

[osma]

I was able to reproduce this with Fuseki 1.3.0. DISTINCT seems to correct the problem.