A number of RDF Stream Processing (RSP) systems have been developed to support processing of streaming Linked Data, however, due to the lack of a standard query language they all provide di erent extensions. The RSP Community Group1 is in the process of standardizing the RSP query language (RSP-QL), which incorporates many features from existing RSP languages. In this paper we present a demo showing how RSP-SPIN, a SPIN extension for RSP-QL, can be used to encapsulate RSP-QL queries as RDF, which can then be used to support serialization into multiple RSP languages. This can reduce the e ort required to produce and maintain queries for RSP benchmarks, since developers can focus on a single representation per query, and assist developers in combining or switching between di erent RSP engines.
The amount of data published as online streams is increasing. While the streams often contain structured or semi-structured data they are often represented using non-standardized vocabularies and models. Semantic Web (SW) technologies have the potential to make this heterogeneity manageable; however, traditional SW technologies are optimized for performance on more or less static data, and do not scale well when dealing with potentially unbounded streams.
RDF Stream Processing (RSP) has been proposed as a way of bridging the
gap between static and streaming Linked Data [
The RSP Community Group2 is working on de ning a standard query
language for RSP (RSP-QL), and the current version of the abstract syntax and
semantics has already been described in some detail [
The diversity of RSP implementations, and their corresponding languages,
makes performance comparisons and switching between engines complicated.
General benchmarks, focusing on di erent aspects of RSP processing, have been
proposed [
In this demo we show how RSP-SPIN, an extension of the SPIN
Modeling Vocabulary4, can be used to represent RSP-QL queries in RDF. The demo
shows how each query in the CSRBench [
Sharing and reusing queries is an important aspect in most contexts where people need to interact with structured data. Many relational databases support some form of stored procedures, which are ready-made parameterized queries that can be instantiated by simply providing the required parameters. Storing queries in such ways has several advantages compared to relying on the users to provide the queries. For example, stored queries dramatically lower the threshold for inexperienced users, can be used to protect against query injections, and to support access control. Stored queries also means that users to not have to have a detailed understanding of the underlying data models, or have any experience in crafting queries themselves.
The SPIN Modeling Vocabulary and the SPIN API was developed in part to provide this functionality for Semantic Web applications. SPIN allows queries to stored as RDF, and enables the representation of parameterized query templates, where query variables can be bound to speci c values at runtime.
RSP-SPIN extends the functionality of SPIN to model RSP-QL. Since RSPQL is based on SPARQL the number of changes made to the base model of SPIN is small, and most changes simply involve adding the appropriate properties, classes, and keywords to support the additional concepts introduced in RSPQL. This means that compatibility with standard SPIN can be maintained.
RSP-SPIN and the RSP-SPIN API are released as open source5. The current version has excluded a some features that have been up for discussion in the RSP Group, which have yet to be discussed in su cient depth. In particular the
excluded features include blank nodes for named graphs, and temporal patterns for Complex Event Processing.
The user interface provided for the demo can be seen in Figure 1. The queries are represented as RSP-SPIN, which can be serialized into any of the supported query languages. Activating the RSP-SPIN view displays the underlying RDF model. The user interface does not support user-de ned parameters for templates, but this is demonstrated in the prede ned queries and on the RSP-SPIN homepage6.
Switching between views displays the current query in di erent formats. The user can pick one of the prede ned queries available in the drop-down menu at the bottom, containing the full set of CSRBench queries. The list also includes a few examples that illustrate the expressivity of RSP-QL and how the CQELS, C-SPARQL, and SPARQLstream serializers handle unsupported RSP-QL features. User de ned queries can be parsed from any view containing a parse button. Pressing the button will pass the text to the web service, which will attempt to create a template for that query. If successful the newly created query will be loaded into the system and all views will be updated accordingly. The live demo is available at http://ontology.ida.liu.se:8680/iswc/rsp-spin/.
Acknowledgments This work was partially supported by the EU FP7 project Visual Analytics for Sense-making in Criminal Intelligence Analysis (VALCRI) under grant number FP7-SEC-2013-608142.