COOL-WD: A Completeness Tool for Wikidata
Fariz Darari, Radityo Eko Prasojo, Simon Razniewski, Werner Nutt
1. Complete data is the new gold!
• Knowledge bases such as Wikidata have
no way of annotating that data is complete
• Consumers may lack trust in data
• Especially relevant for aggregate and negation queries
• “Number of lakes in Zurich”  Is the result correct?
• Editors may not know where to focus
their efforts
• Not clear where data still needs to be added
• “Should one work on Dijkstra prize winners,
on states of Austria, or on lakes in Zurich?”
5. Related publications
• Predicting Completeness in Knowledge Bases.,
L. Galárraga, S. Razniewski, A. Amarilli, F.M. Suchanek, WSDM 2017
• Enabling Fine-Grained RDF Data Completeness Assessment?, F.
Darari, S. Razniewski, R.E. Prasojo, W. Nutt, ICWE 2016
• Managing and Consuming Completeness Information for Wikidata
Using COOL-WD, R. E. Prasojo, F. Darari, S. Razniewski, W. Nutt,
COLD workshop at ISWC 2016
2. Theoretical Background
• Completeness statements have been
proposed as a way to model complete
information [Darari et al., ISWC'13, ICWE'16]
• Assert that for a certain topic,
the knowledge base contains
all true information
• “Complete for districts of Zurich”  All 12
Districts are listed in Wikidata!
3. COOL-WD
• A tool for managing completeness information
for Wikidata
• Allows to assert completeness of predicates
of Wikidata entities
• A user script allows completeness management
directly from Wikidata pages
4. Usage
• COOL-WD contains currently over 10,000
completeness statements
• Completeness statements are used in other
applications including rule mining and
query soundness checking
Visit COOL-WD at http://cool-wd.inf.unibz.it
• Completeness of entities and topics
can be tracked
• SPARQL queries can be analyzed wrt.
the completeness of their results
Annotating complete data lets KB editors and users
know which data is already gold!
So far only a theory construct..
SELECT * WHERE { wd:Q72 wdt:P150 ?district }
// CODE SNIPPET: API request to color property boxes according to their completeness status
$.getJSON( 'https://cool-wd.inf.unibz.it/corner_wd/gadget?callback=?', 'entityId='+entityId, function ( result
) {
...
completeProperties.push("P" + result[i].propertyId);
...
// for every property, do
$( '.wikibase-statementgroupview-property' ).each( function() {
...
// set color
if( $.inArray(propertyId, completeProperties) > -1 ) // if complete then
$this.css( "background-color", colorComplete );
else
$this.css( "background-color", colorIncomplete );
.
.
.

COOL-WD: A Completeness Tool for Wikidata

  • 1.
    COOL-WD: A CompletenessTool for Wikidata Fariz Darari, Radityo Eko Prasojo, Simon Razniewski, Werner Nutt 1. Complete data is the new gold! • Knowledge bases such as Wikidata have no way of annotating that data is complete • Consumers may lack trust in data • Especially relevant for aggregate and negation queries • “Number of lakes in Zurich”  Is the result correct? • Editors may not know where to focus their efforts • Not clear where data still needs to be added • “Should one work on Dijkstra prize winners, on states of Austria, or on lakes in Zurich?” 5. Related publications • Predicting Completeness in Knowledge Bases., L. Galárraga, S. Razniewski, A. Amarilli, F.M. Suchanek, WSDM 2017 • Enabling Fine-Grained RDF Data Completeness Assessment?, F. Darari, S. Razniewski, R.E. Prasojo, W. Nutt, ICWE 2016 • Managing and Consuming Completeness Information for Wikidata Using COOL-WD, R. E. Prasojo, F. Darari, S. Razniewski, W. Nutt, COLD workshop at ISWC 2016 2. Theoretical Background • Completeness statements have been proposed as a way to model complete information [Darari et al., ISWC'13, ICWE'16] • Assert that for a certain topic, the knowledge base contains all true information • “Complete for districts of Zurich”  All 12 Districts are listed in Wikidata! 3. COOL-WD • A tool for managing completeness information for Wikidata • Allows to assert completeness of predicates of Wikidata entities • A user script allows completeness management directly from Wikidata pages 4. Usage • COOL-WD contains currently over 10,000 completeness statements • Completeness statements are used in other applications including rule mining and query soundness checking Visit COOL-WD at http://cool-wd.inf.unibz.it • Completeness of entities and topics can be tracked • SPARQL queries can be analyzed wrt. the completeness of their results Annotating complete data lets KB editors and users know which data is already gold! So far only a theory construct.. SELECT * WHERE { wd:Q72 wdt:P150 ?district } // CODE SNIPPET: API request to color property boxes according to their completeness status $.getJSON( 'https://cool-wd.inf.unibz.it/corner_wd/gadget?callback=?', 'entityId='+entityId, function ( result ) { ... completeProperties.push("P" + result[i].propertyId); ... // for every property, do $( '.wikibase-statementgroupview-property' ).each( function() { ... // set color if( $.inArray(propertyId, completeProperties) > -1 ) // if complete then $this.css( "background-color", colorComplete ); else $this.css( "background-color", colorIncomplete ); . . .