Pipelining e-Entity + e-Link, the second service fails

[andish]

Request:

POST /current/pipelining/chain HTTP/1.1
Host: api.freme-project.eu
Content-Type: application/json

[
    {
      "method": "POST",
  "endpoint": "http://api.freme-project.eu/current/e-entity/freme-ner/documents",
  "parameters": {
    "language": "en",
    "dataset" :"dbpedia"
  },
  "headers": {
    "content-type": "text/plain",
    "accept": "text/turtle"
  },
  "body": "France and Germany came up with a plan"
    },
    {
  "method": "POST",
  "endpoint": "http://api.freme-project.eu/current/e-link/documents/",
  "parameters": {
    "templateid": "2"
  },
  "headers": {
    "content-type": "text/turtle",
    "accept": "text/turtle"
  }
}
  ]

Response

{
  "exception": "eu.freme.common.exception.BadRequestException",
  "path": "/e-link/documents/",
  "message": "It seems your SPARQL template is not correctly defined.",
  "error": "Bad Request",
  "status": 400,
  "timestamp": 1462266259397
}

This error is input-text-specific, some other cases there is no error.

[jnehring]

@ArneBinder please try to reproduce this bug and find out whats wrong.

[ArneBinder]

The problem seems to rely on com.hp.hpl.jena.query. When executing the following e-link request:

curl -X POST -H "Content-Type: text/turtle" -H "Accept: text/turtle" -d '@prefix xsd:   <http://www.w3.org/2001/XMLSchema#> .
@prefix itsrdf: <http://www.w3.org/2005/11/its/rdf#> .
@prefix nif:   <http://persistence.uni-leipzig.org/nlp2rdf/ontologies/nif-core#> .

<http://freme-project.eu/#char=0,38>
        a               nif:RFC5147String , nif:Context , nif:String ;
        nif:beginIndex  "0"^^xsd:int , "0"^^xsd:nonNegativeInteger ;
        nif:endIndex    "38"^^xsd:int , "38"^^xsd:nonNegativeInteger ;
        nif:isString    "France and Germany came up with a plan"^^xsd:string .

<http://freme-project.eu/#char=0,6>
        a                     nif:RFC5147String , nif:Word , nif:String , nif:Phrase ;
        nif:anchorOf          "France"^^xsd:string ;
        nif:beginIndex        "0"^^xsd:int ;
        nif:endIndex          "6"^^xsd:int ;
        nif:referenceContext  <http://freme-project.eu/#char=0,38> ;
        itsrdf:taClassRef     <http://dbpedia.org/ontology/PopulatedPlace> , <http://dbpedia.org/ontology/Place> , <http://dbpedia.org/ontology/Location> , <http://dbpedia.org/ontology/Country> , <http://nerd.eurecom.fr/ontology#Location> ;
        itsrdf:taConfidence   "0.8533452900894164"^^xsd:double ;
        itsrdf:taIdentRef     <http://dbpedia.org/resource/France> .

<http://freme-project.eu/#char=11,18>
        a                     nif:Word , nif:RFC5147String , nif:Phrase , nif:String ;
        nif:anchorOf          "Germany"^^xsd:string ;
        nif:beginIndex        "11"^^xsd:int ;
        nif:endIndex          "18"^^xsd:int ;
        nif:referenceContext  <http://freme-project.eu/#char=0,38> ;
        itsrdf:taClassRef     <http://dbpedia.org/ontology/Place> , <http://nerd.eurecom.fr/ontology#Location> , <http://dbpedia.org/ontology/Country> , <http://dbpedia.org/ontology/PopulatedPlace> , <http://dbpedia.org/ontology/Location> ;
        itsrdf:taConfidence   "0.8074708147491738"^^xsd:double ;
        itsrdf:taIdentRef     <http://dbpedia.org/resource/Germany> .' "http://localhost:8080/e-link/documents?templateid=2"

with a small modification in exception/error passing I get:

{
  "exception": "eu.freme.common.exception.BadRequestException",
  "path": "/e-link/documents",
  "message": "Could not process the enrichment result from the endpoint=http://live.dbpedia.org/sparql executing the query=PREFIX dbpedia: <http://dbpedia.org/resource/> PREFIX dbpedia-owl: <http://dbpedia.org/ontology/> PREFIX geo: <http://www.w3.org/2003/01/geo/wgs84_pos#> CONSTRUCT { ?event <http://dbpedia.org/ontology/place> <http://dbpedia.org/resource/Germany> . } WHERE { ?event <http://dbpedia.org/ontology/place> <http://dbpedia.org/resource/Germany> .  } LIMIT 10. Error message: [line: 3, col: 19] Unknown char: –(8211;0x2013)",
  "error": "Bad Request",
  "status": 400,
  "timestamp": 1462814026872
}

Calling the referenced dbpedia sparql endpoint with the query mentioned in the exception gives this result:

@prefix dbo:    <http://dbpedia.org/ontology/> .
@prefix dbr:    <http://dbpedia.org/resource/> .
dbr:Battle_of_Jena–Auerstedt  dbo:place   dbr:Germany .
dbr:Battle_of_Maxen dbo:place   dbr:Germany .
dbr:Battle_of_Meissen   dbo:place   dbr:Germany .
dbr:Battle_of_Minden    dbo:place   dbr:Germany .
dbr:Battle_of_Torgau    dbo:place   dbr:Germany .
dbr:Battle_of_Friedlingen   dbo:place   dbr:Germany .
dbr:Saxon_Wars  dbo:place   dbr:Germany .
<http://dbpedia.org/resource/Battle_of_Berlin_(RAF_campaign)>   dbo:place   dbr:Germany .
dbr:Siege_of_Weinsberg  dbo:place   dbr:Germany .
<http://dbpedia.org/resource/Battle_of_the_Heligoland_Bight_(1939)> dbo:place   dbr:Germany .

Trying to read the char – of the uri dbr:Battle_of_Jena–Auerstedt in the function com.hp.hpl.jena.sparql.engine.http.QueryEngineHTTP:execModel(Model model) leads to the exception.

[jnehring]

@m1ci the bug originates from e-link. Please take a look at the bug.

[m1ci]

The issue is within Jena v 2.11.2 with the – char which is considered as invalid. We should consider upgrading Jena.
I found Jena dependency in FREMECommon
https://github.com/freme-project/FREMECommon/blob/master/pom.xml
@jnehring Where else is Jena set as dependency?

[jnehring]

I think that Jena dependency is only defined in FREMECommon. I will update it to the latest version.

[m1ci]

please upgrade to Jena 2.13.0
This Jena version is supported by Linked Data Fragments

[jnehring]

@ArneBinder please update FREMECommon Jena dependency to Jena 2.13.0 and check if the bug is fixed by that.

[jnehring]

We tried the upgrade to Jena 2.13.0 and it did not solve the problem. We cannot upgrade to higher versions of Jena because the LDF client is not compatible to higher Jena versions.

So we cannot fix this bug.

[andish]

See 6.4 Escape Sequences
at https://www.w3.org/TR/turtle/#reserved

Seems that in turtle, the IRI cannot contain -.

I asked that DBpedia endpoint - it returns more than 13 000 resource IRIs containg character -. So i guest than this error will come up regularly.

So one solution would be asking the endpoint for other rdf syntax, e.g triples. [in case if e-link asks n3/turtle from endpoints].

[jnehring]

great idea! thanks @andish for the research and the proposal.
we will discuss this at next developers call.

[borriellom]

Any development on this issue?

[jnehring]

No new developments. I put this issue on the agenda of next developers call.

[jnehring]

On todays dev call we agreed that @sandroacoelho and @m1ci will work on the issue

[m1ci]

the issue is fixed, @andish @borriellom please check!

[borriellom]

It's fixed. Thank you!

[jnehring]

Fixed on freme-live through the release