miguel bento alves, joão ferreira nunes · 2018. 11. 12. · instead of using fuseki (the sparql...

Miguel Bento Alves, João Ferreira Nunes

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

Our project arises with the need to share andmake public the data produced under theTREASURE project - a Research & InnovationAction financed by European Commission underthe Horizon 2020 (grant agreement no. 634476).

The aim of the project is to improve knowledge,skills and competences necessary to develop newsustainable pork chains based on European localpig genetic resources (local breeds).

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

Our goal consists in publishing TREASUREdata by means of an Open Data approach.

The data will be available for all stakeholdersin a standard format.

Furthermore, Open Data implies the use ofstandards such as HTTP, RDF or SPARQL,making it easier to use on the web.

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

Open Data's principle claims that data shouldbe freely available, without any kind ofrestrictions from copyright, patents or othermechanisms of control.

Another key concept that it is implicit to thisideal, is the interoperability, which refers tothe capability of several systems andorganizations in working together.◦ In this specific case, it refers to the capability to

combine - or inter-operate - different sets of data.

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

Initially, the information requirements wereanalyzed based on a relational modelapproach to create a relational database.

In order to reuse all the work producedduring the initial phase, it was decided toreplicate the relational model for a SemanticWeb approach.

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

All data will be kept as RDF;

In our Semantic Web approach, all relationalrules are guaranteed in the RDF data;

Although the focus of our work was topublish the produced data from theTREASURE project on an Open Data approach,the developed system that we designed isadaptable to any relational database.

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

We proposed a three layers model, where:

◦ at the upper-level the most important (or mostused) concepts of the relational model are modeled;

◦ at the middle-level the meta-model of therelational database is modeled;

◦ at the lower-level the database information isrepresented.

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

With the two highest layers, we have all theknowledge on how the information in adatabase is organized, and from that we canextract information about what is modeled;

This allows us to support reasoning on thedata model;

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

In the upper-level, we have created an ontologyto represent the concepts of relationaldatabases.

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

At the middle-level it is represented themeta-model of the database, namely whichtables were created, which fields have eachtable, what are the primary keys of the tablesand the foreign keys.

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

Can be fed by thedatabasecatalogue (datadictionary)

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

In the most specific layer (the lower-level),the data itself is represented.

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

We’ve developed two different tools to manipulatedata in the open data repository.

The first one allows to select one local database andtransfer the data to the Open Data repository.

The second one is a SPARQL endpoint, that can beused either in a program or with a Web interface, thatallows the execution of SPARQL commands in thecentral repository.

Both tools guarantees the integrity of the dataconsidering the relational constraints implemented.

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

We’ve developed our own SPARQL endpointinstead of using Fuseki (the SPARQL server ofJena package), because we implementedseveral relational constraints over our centralrepository and we wanted to control theintegrity of the data against the relationalconstraints.

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

Our system was developed using the JenaFramework, a free and open source Javaframework for building Semantic Webapplications.

Jena provides a programmatic environment forRDF, RDFS, OWL, a query engine for SPARQL andit includes a rule-based inference engine.

Jena is widely accepted for Semantic Webapplications because it offers an "all-in-one" Javasolution.

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

All relational constraints are guaranteed bysemantic rules:

RULE 1: avoid repeated fields in a table;

RULE 2: avoid violation of primary key constraint;

RULE 3: avoid violation of foreign key constraint;

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

We’ve developed a system, in a semantic webapproach, to publish relational data as OpenData;

A three layer model was proposed to supportthe knowledge about relational data;◦ A relational model ontology it was also developed;

Semantic rules were developed to support therelational data constraints;

Miguel Bento Alves, João Ferreira Nunes

An Open Data Approach to Publish Relational Data

We’ve developed two tools to manipulate datain the open data repository.

◦ The first one allows to select one local databaseand transfer the data to the Open Data repository.

◦ The second one is a SPARQL endpoint, that can beused either in a program or with a Web interface,that allows the execution of SPARQL commands inthe central repository.

Miguel Bento Alves, João Ferreira Nunes

THANK YOU