=Paper=
{{Paper
|id=Vol-2180/paper-74
|storemode=property
|title=Demonstration of Comunica, a Web framework for Querying Heterogeneous Linked Data
Interfaces
|pdfUrl=https://ceur-ws.org/Vol-2180/paper-74.pdf
|volume=Vol-2180
|authors=Joachim Van Herwegen,Ruben Taelman,Miel Vander Sande,Ruben Verborgh
|dblpUrl=https://dblp.org/rec/conf/semweb/HerwegenTSV18
}}
==Demonstration of Comunica, a Web framework for Querying Heterogeneous Linked Data
Interfaces==
https://ceur-ws.org/Vol-2180/paper-74.pdf
Demonstration of Comunica, a Web framework for
querying heterogeneous Linked Data interfaces
Joachim Van Herwegen, Ruben Taelman, Miel Vander Sande, Ruben Verborgh
IDLab, Department of Electronics and Information Systems, Ghent University – imec
Abstract. Linked Data sources can appear in a variety of forms, going from
SPARQL endpoints to Triple Pattern Fragments and data dumps. This
heterogeneity among Linked Data sources creates an added layer of complexity
when querying or combining results from those sources. To ease this problem, we
created a modular engine, Comunica, which has modules for evaluating SPARQL
queries and supports heterogeneous interfaces. Other modules for other query or
source types can easily be added. In this paper we showcase a Web client that
uses Comunica to evaluate federated SPARQL queries through automatic source
type identification and interaction.
1. Introduction
There are a multitude of ways we can access Linked Data these days. Some of the
more commonly used ones are SPARQL endpoints [1], Triple Pattern Fragments
(TPF) [2] and its variations [3, 4], Linked Data documents [5] and data dumps. These
all have their own methods on how they can be accessed and help solve SPARQL
queries [6]. While a SPARQL endpoint can execute queries on its own and can require
a significant amount of server effort, data dumps will require client-side processing to
produce more granular results and is less intensive for servers. This trade-off is
measured as client cost on the Linked Data Fragments axis [2].
Having all these heterogeneous interfaces greatly complicates federated queries.
While resolving such a query, different actions have to be taken depending on the
source that is being accessed. Different solutions might also be required depending on
the combination of sources. In case of a single SPARQL endpoint, a single query will
suffice. On the other hand, if all sources are data dumps they all have to be
downloaded and parsed client-side. But what if some sources are SPARQL endpoints
and some are data dumps?
To this end we created a modular Linked Data client called Comunica [7]. In our
ISWC 2018 article we describe how this client can easily be extended to support a
variety of sources and algorithms. This allows everyone to quickly set up a federated
SPARQL client without having to worry about the sources, and to easily extend it
should more types be required.
In this article we describe how we will showcase the heterogeneous features of
Comunica. We created a Web client capable of executing federated SPARQL queries
over heterogeneous interfaces, as an extension of the Triple Pattern Fragments Web
client [8]. Additionally, this client will automatically identify the type of all sources.
This way the end-user only has to provide the URLs and query; Comunica will take
care of the rest.
2. Comunica
Comunica [7] is a modular meta engine that enables the instantiation of specific
engines with their functionality described by modules through semantic configuration
files. We released 80+ modules (https://github.com/comunica/comunica) that can be
combined to fully replicate all features of the original TPF client (https://github.com/
LinkedDataFragments/Client.js). Comunica is not limited to simply solving SPARQL
queries: new modules can be added to solve new problems and add additional
features. Similarly, existing functionality can easily be switched out for others to
quickly compare different implementations and ideas.
Every module consists of two parts: the source code and its semantic description.
The second part is a collection of Linked Data documents describing the functionality
of the corresponding module. These are used by the Components.js [9] dependency
injection framework to instantiate and link all modules together.
The current collection of Comunica modules offers more functionality than the
original TPF client. The default configuration allows users to query different kinds of
Linked Data besides TPF interfaces. We also provide support for SPARQL endpoints,
Linked Data documents and HDT [10] files. These can all be combined in a single
federated query by making use of the federated TPF algorithm and utilizing
Comunica modules to allow triple pattern queries on those different source types.
3. Demonstration overview
In this demonstration, we offer the possibility for executing SPARQL queries over
a federation of heterogeneous interfaces. This demonstration can be used directly
within the browser, and is available on the Comunica website (http://
comunica.linkeddatafragments.org/).
This demonstration is an adaption of the Triple Pattern Fragments Web client [8],
with the main difference that instead of using the Triple Pattern Fragments engine for
querying, it uses the Comunica engine. The implementation of this Web client is
available on GitHub (https://github.com/comunica/jQuery-Widget.js), under the open
MIT license so that it can be reused for different use cases.
We provide a collection of example queries with a predefined set of sources, where
some queries federate over different heterogeneous sources. Fig. 1 shows an example
query that federates over a Triple Pattern Fragments interface and a Linked Data
document. Additionally, users can also write custom queries, and add more
datasources by their URL.
Fig. 1: Example SPARQL query in the Comunica Web client that federates over the
DBpedia Triple Pattern Fragments interface and a FOAF profile.
At the time of writing, SPARQL endpoints, Triple Pattern Fragments interfaces and
raw RDF files can be queried. Internally, Comunica will identify the source type
through a set of heuristics. SPARQL endpoints are tested using a simple ASK query
through the SPARQL protocol [1]. Triple Pattern Fragments interfaces are tested by
checking if the required set of hypermedia controls is available. Finally, RDF files are
tested with the lowest priority by checking their content type.
References
1. Feigenbaum, L., Todd Williams, G., Grant Clark, K., Torres, E.: SPARQL 1.1
Protocol. W3C, https://www.w3.org/TR/2013/REC-sparql11-protocol-20130321/
(2013).
2. Verborgh, R., Vander Sande, M., Hartig, O., Van Herwegen, J., De Vocht, L., De
Meester, B., Haesendonck, G., Colpaert, P.: Triple Pattern Fragments: a Low-cost
Knowledge Graph Interface for the Web. Journal of Web Semantics. 37–38,
(2016).
3. Hartig, O., Buil-Aranda, C.: Bindings-Restricted Triple Pattern Fragments. In:
Proceedings of the 15th International Conference on Ontologies, DataBases, and
Applications of Semantics. pp. 762–779 (2016).
4. Taelman, R., Vander Sande, M., Verborgh, R., Mannens, E.: Versioned Triple
Pattern Fragments: A Low-cost Linked Data Interface Feature for Web Archives.
In: Proceedings of the 3rd Workshop on Managing the Evolution and Preservation
of the Data Web (2017).
5. Berners-Lee, T.: Linked Data. https://www.w3.org/DesignIssues/LinkedData.html
(2009).
6. Harris, S., Seaborne, A., Prud’hommeaux, E.: SPARQL 1.1 Query Language.
W3C, https://www.w3.org/TR/2013/REC-sparql11-query-20130321/ (2013).
7. Taelman, R., Van Herwegen, J., Vander Sande, M., Verborgh, R.: Comunica: a
Modular SPARQL Query Engine for the Web. In: Proceedings of the 17th
International Semantic Web Conference (2018).
8. Verborgh, R., Hartig, O., De Meester, B., Haesendonck, G., De Vocht, L., Vander
Sande, M., Cyganiak, R., Colpaert, P., Mannens, E., Van de Walle, R.: Low-Cost
Queryable Linked Data through Triple Pattern Fragments. In: International
Semantic Web Conference (Posters & Demos). pp. 13–16 (2014).
9. Taelman, R.: Components.js. http://componentsjs.readthedocs.io/en/latest/
10. Fernández, J.D., Martínez-Prieto, M.A., Gutiérrez, C., Polleres, A., Arias, M.:
Binary RDF Representation for Publication and Exchange (HDT). Web
Semantics: Science, Services and Agents on the World Wide Web. 19, 22–41
(2013).