Event News

Semantic Web Seminar II


October 27 (Thursday), 2016
National Institute of Informatics
12F, Room 1208+1210
Executing SPARQL queries over Mapped Document Stores with SparqlMap-M
Jorg Unbehauen
AKSW, Leipzig University

With the increasing adoption of NoSQL data base systemslike MongoDB or CouchDB more and more applications storestructured data according to a non-relational, document orientedmodel. Exposing this structured data as Linked Data is currentlyinhibited by a lack of standards as well as tools and requires theimplementation of custom solutions. While recent efforts aim atexpressing transformations of such data models into RDF in astandardized manner, there is a lack of approaches whichfacilitate SPARQL execution over mapped non-relational datasources. With SparqlMap-M we show how dynamic SPARQL access tonon-relational data can be achieved. SparqlMap-M is an extensionto our SPARQL-to-SQL rewriter SparqlMap that performs a (partial)transformation of SPARQL queries by using a relational abstractionover a document store. Further, duplicate data in the documentstore is used to reduce the number of joins and customoptimizations are introduced. Our showcase scenario employs theBerlin SPARQL Benchmark (BSBM) with different adaptions to adocument data model. We use this scenario to demonstrate theviability of our approach and compare it to different MongoDBsetups and native SQL.

Simplified RDB2RDF Mapping
Claus Stadler
AKSW, Leipzig University

The combination of the advantages of widely usedrelational databases and semantic technologies has attractedsignificant research over the past decade. In particular, mappinglanguages for the conversion of databases to RDF knowledge baseshave been developed and standardized in the form of R2RML. In thisarticle, we first review those mapping languages and then devisework towards a unified formal model for them. Based on this, wepresent the Sparqlification Mapping Language (SML), which providesan intuitive way to declare mappings based on SQL VIEWS and SPARQLconstruct queries. We show that SML has the same expressivity asR2RML by enumerating the language features and show thecorrespondences, and we outline how one syntax can be convertedinto the other. A conducted user study for this paper juxtaposingSML and R2RML provides evidence that SML is a more compact syntaxwhich is easier to understand and read and thus lowers the barrierto offer SPARQL access to relational databases.


D2RQ Mapper: Accelerating RDFization in the Life Science Domain
Yasunori Yamamoto
Database Center for Life Science

D2RQ Mapper is a web application to edit a mapping file of D2RQ, a middleware to bridge Relational Database (RDB) and Resource Description Framework (RDF). D2RQ Mapper supports to export a mapping file in the D2RQ mapping language or the R2RML formats. A D2RQ mapping file defines how to map data stored in an RDB to RDF in the turtle format, and to write it by a text editor is cumbersome. D2RQ Mapper assists you to edit it by contextualizing input forms in the mapping language. We provide a Docker image of D2RQ Mapper, and so you can easily use it within your intranet.http://d2rq.dbcls.jp/

Test-driven Evaluation of Linked Data Quality
Sebastian Hellmann
AKSW & KILT Competence Center, Institute for Applied Informatics (InfAI)

Linked Open Data (LOD) comprises of an unprecedentedvolume of structured data on the Web. However, these datasets areof varying quality ranging from extensively curated datasets tocrowd-sourced or extracted data of often relatively lowquality. We present a methodology for test-driven qualityassessment of Linked Data, which is inspired by test-drivensoftware development. We argue, that vocabularies, ontologies andknowledge bases should be accompanied by a number of test-cases,which help to ensure a basic level of quality. We present amethodology for assessing the quality of linked data resources,based on a formalization of bad smells and data qualityproblems. Our formalization employs SPARQL query templates, whichare instantiated into concrete quality test queries. Based on anextensive survey, we compile a comprehensive library of dataquality test patterns. We perform automatic test instantiationbased on schema constraints or semi-automatically enrichedschemata and allow the user to generate specific testinstantiations that are applicable to a schema or dataset. Weprovide an extensive evaluation of five LOD datasets, manual testinstantiation for five schemas and automatic test instantiationsfor all available schemata registered with LOV. One of the mainadvantages of our approach is that domain specific semantics canbe encoded in the data quality test cases, thus being able todiscover data quality problems beyond conventional qualityheuristics.


RDF Editing on the Web with REX
Claus Stadler
AKSW, Leipzig University

While several tools for simplifying the task ofvisualizing (SPARQL accessible) RDF data on the Web are availabletoday, there is a lack of corresponding tools for exploitingstandard HTML forms directly for RDF editing. The few relatedexisting systems roughly fall in the categories of (a)applications that are not aimed at being reused as components, (b)form generators, which automatically create forms from a givenschema -- possibly derived from instance data -- or (c) formtemplate processors which create forms from a manually createdspecification. Furthermore, these systems usually come with theirown widget library, which can only be extended by wrappingexisting widgets. In this paper, we present the AngularJS-basedRdf Edit eXtension (REX) system, which facilitates the enhancementof standard HTML forms as well as many existing AngularJS widgetswith RDF editing support by means of a set of HTML attributes. Wedemonstrate our system though the realization of several usagescenarios.

Sebastian Hellmann
KSW & KILT Competence Center, Institute for Applied Informatics (InfAI)

The Natural Language Processing Interchange Format (NIF)is an RDF/OWL-based format that aims to achieve interoperabilitybetween Natural Language Processing (NLP) tools, languageresources and annotations. The introduction will show its usage inthe FREME project:http://www.freme-project.eu/

Challenges for DBpedia
Sebastian Hellmann
KSW & KILT Competence Center, Institute for Applied Informatics (InfAI)

The talk will summarize the current state of DBpedia andthen go into the technical challenges that DBpedia and itscommunity are facing in the future. This includes: Extraction ofarticle text and NLP, Extending DBpedia with further facts,Wikidata, Identifier management, Metadata problems, andContributing links to DBpedia.

Application examples of DBpedia Japanese
Fumihiro Kato
National Institute of Informatics

DBpedia Japanese is getting used widely in JapaneseLinked Open Data community since 2012. This talk will introduceapplications using DBpedia Japanese.