Wikidata as a collaborative Semantic Web data source has enjoyed increasing prominence, storing over 150 million facts about more than 26 million entities. Yet it is missing a way to assess its completeness. In this demo we present COOL-WD, a tool for supporting the completeness lifecycle of Wikidata, that is, the creation, view, update, and consumption of metadata about Wikidata completeness. COOL-WD is available at http://cool-wd.inf.unibz.it/ and has so far collected more than 10,000 completeness statements.
Completeness Lifecycle. When talking about completeness statements of RDF data sources, we conceive that the statements go through a lifecycle consisting of four phases as shown in Fig. 1:
1. Creation: Completeness statements about RDF data sources are created, where provenance (e.g., author, reference, timestamp) of the statements can also be added.
2. View : Completeness statements (and their provenance), as well as parts of data captured by the statements, are available for viewing. 3. Update: Completeness statements can be updated (i.e., edited or deleted), if, for example, they are no longer valid or incorrectly given.
4. Consumption: Various consumption tasks such as query completeness checking and completeness analytics are performed.
The stages are collectively referred to as cycle because the consumption stage might reveal (in-)completeness problems of the data source, thus triggering the further creation of completeness statements.
Outcome. The intended outcome of the demo session is three-fold, that visitors shall become aware of: (i) how natural completeness information is for KBs (e.g., Wikidata); (ii) how the completeness lifecycle models a process of creating, viewing, updating, and consuming completeness information; and (iii) how COOL-WD realizes support for the completeness lifecycle of Wikidata.
While the present demonstration focuses on Wikidata, the need for expressing
and storing completeness information is not limited to this KB. Our goal is to
increase awareness of the completeness issue on the Semantic Web, and the need
for considering it in the design of future methodologies, standards, and KBs.
Related Work. Data completeness relates to the breadth, depth, and scope of
information in the data [
COOL-WD is in active use by the Wikidata community and contains so far around 10,000 completeness statements, available for bulk download at http: //completeness.inf.unibz.it/rdf-export/. 3
On the Wikidata page of Austria4, the user is left clueless whether all the states of Austria are present. Moreover, it is unknown which of the states are already complete for their districts, hindering where to put focus on data collection about Austria. 1. Creation. By comparing information in Wikidata with the o cial website of Austria,5 users can observe that Wikidata is complete for the states. Thus the user can add the corresponding completeness statement, as well as the reference URL. Using the direct editing, the adding is as simple as clicking the respective property to the Wikidata page about Austria. The Wikidata username and timestamp are added accordingly. 2. View. As illustrated in Fig. 2 (a), the user can view completeness statements about Austria, as well as the provenance, by clicking the \(i)" icon. Greenhiglighted properties indicate completeness, while yellow indicates unknown. For instance, the user can see that the completeness of short names is unknown. We also provide a Linked Data API for COOL-WD completeness statements. For example, the dereferenceable URI for the statement (Q40,P150) as above is http://cool-wd.inf.unibz.it/resource/statement-Q40-P150. There, the user can view an RDF description of the completeness statement, modeled using our completeness vocabulary (http://completeness.inf.unibz.it/sp-vocab) and the W3C PROV vocabulary (http://www.w3.org/ns/prov). 3. Update. The user may notice a completeness statement for \diplomatic relation", which is not valid anymore, as Austria in the meantime has also established diplomatic relations with Kyrgyzstan6. The user can therefore remove the 3 http://www.gwtproject.org/ 4 https://www.wikidata.org/wiki/Q40 5 https://www.parlament.gv.at/ENGL/PERK/BOE/ 6 https://www.embassypages.com/missions/embassy11893 statement simply by clicking the green box. To update the URL of an existing statement, the user can edit the reference eld after clicking \(i)". 4. Consumption. So far, COOL-WD supports three consumption tasks: data completeness tracking, completeness analytics, and query completeness diagnostics. The tracking shows the progress of completing the data of a single entity in terms of the fraction of the (known) non-functional properties that are complete. The completeness analytics summarizes in a tabular view the completeness of a class of entities. For example, for the class of Austrian states, the user might be interested in the completeness percentage of their districts and borders (i.e., how complete is the class for these properties?). The percentage is computed based on whether completeness statements exist or not for the properties of interest wrt. all entities of the class. Finally, query completeness diagnostics enables the completeness assessment of SPARQL basic graph pattern queries (as in Fig. 2 (b)). Additionally, statements used to guarantee query completeness can be shown. Acknowledgements This work was partially funded by the Free University of Bolzano under the TaDaQua project.