The awareness of quality issues in Linked Data is constantly rising as new datasets and applications that consume Linked Data are emerging. In this paper we summarize key problems of Linked Data quality that data consumers are facing and propose approaches to tackle these problems. The majority of challenges presented here have been collected in a Lightning Talk Session at the First Workshop on Linked Data Quality (LDQ2014). A consumer's perception of data quality is highly individual and strongly depends on the eld of application. Therefore, data quality is often regarded as tness for use, e. g. the DBpedia dataset might be appropriate for a simple end-user application but should not be used in critical applications such as the medical domain for treatment decisions. However, quality is a key to the success of the data web and a major barrier for further industry adoption.
In this paper we want to highlight contemporary quality problems that occur in Linked Data and that already are or need to be addressed in future. Likewise, we want to suggest solutions that have been developed in order to tackle these di culties.
Although quality is a commonly used term in Linked Data, it's de nition is far from straightforward. The reason is that Linked Data quality can have di erent meaning for di erent people and in di erent contexts. During the First Workshop on Linked Data Quality (LDQ2014)1, a discussion session was held where people from di erent backgrounds raised their personal thoughts on Linked Data Quality 2. It was surprising to notice the variety of de nitions and concerns that among others included stalled data, version management, changeset updates, RDF typo identi cation, and proper ontology modeling.
RDF validation is a core part of Linked Data quality but
validation alone cannot solve the quality problem. Quality
is tness for use, thus a general methodology [
RDF version management is an additional quality issue that is not natively covered in the Semantic Web technology stack and can facilitate error provenance and tracking. The nondeterministic statement order and blank nodes make graph comparison equivalent to the graph isomorphism problem and thus, beyond polynomial time computation complexity3.
Reusing popular vocabularies or manually creating a correct ontology model can also be seen as a general data quality issue. General purpose vocabularies such as foaf4, skos5, schema.org6 or dbpedia ontology7 usually re ect a swallow depiction of the real world. For many people for example, the dbo:Actor class is not correct since a profession is a role in a person's life and a person can have many di erent roles at di erent stages of his life, e. g. student or spouse. In the 1http://ldq.semanticmultimedia.org/ co-located with 10th SEMANTiCS conference on September 2nd, 2014 in Leipzig, Germany 2http://tinyurl.com/LDQ14LightningTalks 3http://mathworld.wolfram.com/GraphIsomorphism. html 4http://xmlns.com/foaf/0.1/ 5http://www.w3.org/2004/02/skos/core# 6http://schema.org/ 7http://dbpedia.org/ontology/, pre xed dbo: end it depends on the granularity level one wants to re ect in his data but granularity usually comes at the cost of data integration.
Speci c solutions for tackling the problem of Linked Data Quality as a whole are currently far from reality. Nevertheless, in the following subsection we provide an overview of existing work and possible future directions to cope with Linked Data Quality.
Validation is a core part in the quality assessment of Linked
Data. Although RDF exists already for many years there
exists no o cial standard for Linked Data validation at the
time of writing and a W3C working group has just been
formed to de ne one8. Existing Linked Data application
can either rely on ad-hoc options or use independently
dened solutions such as: RDF Data Shapes [
However, validation alone cannot be adequate. A general
assessment methodology has to be built around validation
that can interpret the validation results and assess the
quality of the data. Rula et al. [
Data curation in general is a costly process for the publisher.
The distributed nature of Linked (Open) Data may demand
the involvement of multiple data providers to achieve
satisfactory results. On the other hand those, who su er from
low data quality and most often identify these issues, are
the data consumers. Unfortunately, only in some cases they
provide feedback to create awareness of particular problems.
A typical approach for Linked Data consumers is to
duplicate a dataset and x relevant problems within the local
copy. These e orts are rarely communicated and hence not
imitated on the original data so that other consumers could
bene t equally. The Patch Request vocabulary [
The correction of errors in Linked Data should be distributed onto the shoulders of many and possibilities to distribute such changes should be researched. 3.3
8http://www.w3.org/blog/data/2014/09/30/ data-shapes-working-group-launched/ Linked Data is primarily made for machine interpretation and therefore it needs to comply the technical standards. Typical RDF parser implementations do not cope with {even minor{ syntactical errors. Many publishers create RDF data using scripts or perform changes manually. Such modalities raise the risk of introducing syntactical errors, which can be avoided by using RDF tools and programming libraries, such as the Redland RDF API9 and Apache Jena10. Optionally, generated RDF data should be checked by subsequent syntax validation prior to publication with appropriate tools, such as the Raptor RDF parser utility11 and Apache Jena CLI tools12.
Hogan et al. [
Beyond validating the syntax of their RDF serialization,
data providers should also keep an eye on the correct usage
of vocabularies. RDFUnit [
To ensure that entities within a dataset are described in a
form that is required for usage by a particular application,
such structures can be de ned with RDF Data Shapes as it
has been done to the WebIndex data portal [
Continuous updates to and curation of datasets raises the
aim for tracking changes in Linked Data. There is rare
support in Linked Data publishing for versioning as well as for
provenance of changes. Apache Marmotta13 is one Linked
Data publishing platform that supports versioning. R43ples
[
In this paper we have discussed key issues of Linked Data quality aspects and possible ways to tackle them. We see quality as a core {and grey{ component of the semantic web stack that, if addressed correctly and systematically, will enable further adoption. 9http://librdf.org/ 10https://jena.apache.org/ 11http://librdf.org/raptor/rapper.html 12https://jena.apache.org/documentation/io/ #command-line-tools 13http://marmotta.apache.org/
We would like to thank all attendees of the 1st Linked Data Quality Workshop (LDQ2014), especially the people contributing to the Lightning Talks Session. These discussions delivered valuable input and witnessed that this topic is of high importance to researchers and practitioners working with Linked Data. In addition we would also like to thank all reviewers that have helped us and the authors with their valuable comments and feedback.
Maribel Acosta { Karlsruhe Institute of Technology, AIFB, Germany Volha Bryl { University of Mannheim, Germany
Christoph Lange { University of Bonn, Fraunhofer IAIS, Germany Maristella Matera { Politecnico di Milano, Italy Felix Naumann { Hasso Plattner Institute, Germany Matteo Palmonari { University of Milan-Bicocca, Italy Adrian Paschke { Free University of Berlin, Germany Heiko Paulheim { University of Mannheim, Germany Mariano Rico { Universidad Politecnica de Madrid, Spain Anisa Rula { Universita di Milano-Bicocca, Italy Elena Simperl { University of Southampton, United Kingdom Patrick Westphal { AKSW, University of Leipzig, Germany Amrapali Zaveri { AKSW, University of Leipzig, Germany Jun Zhao { Lancaster University, United Kingdom Antoine Zimmermann { ISCOD / LSTI { Ecole Nationale Superieure des Mines de Saint-Etienne, France
Behshid Behkamal { Ferdowsi University of Mashhad, Iran Jeremy Debattista { University of Bonn, Germany Riccardo Del Gratta { Istituto di linguistica Computazionale CNR, Italy