Archive for the ‘R2RML’ Category

R2RML: RDB to RDF Mapping Language

Tuesday, August 14th, 2012

R2RML: RDB to RDF Mapping Language from the RDB2RDF Working Group.

From the news:

This document describes R2RML, a language for expressing customized mappings from relational databases to RDF datasets. Such mappings provide the ability to view existing relational data in the RDF data model, expressed in a structure and target vocabulary of the mapping author’s choice. R2RML mappings are themselves RDF graphs and written down in Turtle syntax. R2RML enables different types of mapping implementations. Processors could, for example, offer a virtual SPARQL endpoint over the mapped relational data, or generate RDF dumps, or offer a Linked Data interface. Comments are welcome through 15 September. (emphasis added)

Subscribe (prior to commenting).

Comments to:

Exposing Databases…

Friday, July 8th, 2011

Exposing databases as linked data using clojure, compojure and R2RML

From the post:

Some of these problems could be solved using some of the technologies developed as part of the semantic web initiative in the past decade. People have started referring to this pragmatic approach to the semantic web with a new title: Linked Data. The pragmatic approach here means putting less emphasis in the inference and ontological layers of the semantic web and just focusing in offer a simple way to expose data in the web linking resources across different web application and data sources.

Many interesting technologies are being developed under the linked data monicker or are commonly associated to it, RDFa for instance. Another of these technologies is R2RML: RDB to RDF Mapping Language.

R2RML describes a standard vocabulary to lift relational data to a RDF graph. It also provides a standard mapping for the relational data. This RDF graph can be serialized to some transport format: RDFa, Turtle, XML and then retrieved by the client. The client can store the triples in the graph locally and use the standard query language SPARQL to retrieve the information. Data from different applications using the same vocabulary (FOAF, GoodRelations) can be easily mixed and manipulated by the client in the same triple store. Furthermore, links to other resources can be inserted inside the RDF graph leading to the discovery of additional information.

Exposing data, even as an RDF graph, is a good thing.

Discovering additional information is also a good thing.

But they fall short of specifying the semantics of data, which is necessary to enable reliable identification of data with the same semantics.

Translating SPARQL queries into SQL using R2RML

Monday, July 4th, 2011

Translating SPARQL queries into SQL using R2RML

From the post:

The efficient translation of SPARQL into SQL is an active field of research in the academy and in the industry. In fact, a number of triple stores are built as a layer on top of a relational solution. Support for SPARQL in these RDF stores supposes the translation of the SPARQL query to a SQL query that can be executed in a certain relational schema.

Some foundational papers in the field include “A Relational Algebra for SPARQL” by Richard Cyganiak that translates the semantics of SPARQL as they were finally defined by the W3C to the Relational Algebra semantics or “Semantics preserving SPARQL-to-SQL translation” by Chebotko, Lu and Fotohui, that introduces an algorithm to translate SPARQL queries to SQL queries.

This latter paper is specially interesting because the translation mechanism is parametric on the underlying relational schema. This makes possible to adapt their translation mechanism to any relational database using a couple of mapping functions, alpha and beta, that map a triple pattern of the SPARQL query and a triple pattern and a position in the triple to a table and a column in the database.

Provided that R2RML offers a generic mechanism for the description of relational databases, in order to support SPARQL queries in any R2RML RDF graph, we just need to find an algorithm that receives as an input the R2RML mapping and builds the mapping functions required by Chebotko et alter algorithm.

The straightest way to accomplished that is using the R2RML mapping to generate a virtual table with a single relation with only subject, predicate and object. The mapping for this table is trivial. A possible implementation of this algorithm can be found in the following Clojure code. (I added links to the Cyganiak and Chebotko papers.)

I recommend this post, as well as the Cyganiak and Chebotko papers to anyone interested in TMQL as background reading. Other suggestions?