The original Semantic Web scenario was proposed to embed HTML pages with RDF metadata inorder to enable search engines to harvest and reason about the web (as structured and semantically enabled information). This scenario was not successful at least because scalability matters. We believe SPARQL is bringing new and different scenarios for using RDF, such as semantic interoperability, data integration, information search and retrieval (rather than the typical web searching), to name few. Organizations can share and exchange data meaningfully, as well as query each other’s data using RDF and SPARQL. One may imagine Amazon receives (or has access to) book catalogues from 1000 publishers in RDF format; integrating this RDF data can be done in few SPARQL queries.
I believe SPARQL has made the web as a database, where each RDF resource is seen as a table; all resources can be accessed and queried using SPARQL, as simple as querying tables using SQL. Imagine a researcher would like to generate his list of publications from the DLPB, Google Scholar, and CiteSeer. Suppose these scholarly libraries generate RDF of the results they provide, or a third service converts their results into RDF. Integrating and mashing up this content can be done easily using SPARQL, I shall illustrate this examples, in later posts.
It is important noting that Oracle 10g and 11g support RDF storage and querying. Oracle stores RDF triples in one simple table, in addition to some indices. Querying RDF in Oracle is done in a SPARQL style. As shown by Oracle, this implementation is scalable. For example, a query (with medium size complexity) over 80 Million RDF triples (5.2 GB) takes less than a second. It has been shown that retrieval cost per result row remains almost the same as the dataset size changes. I believe this simple and scalable support of the RDF technology (by Oracle) will accelerate the adoption of RDF as the web metadata format, and thus SPARQL as the web query language.
I believe SPARQL has made the web as a database, where each RDF resource is seen as a table; all resources can be accessed and queried using SPARQL, as simple as querying tables using SQL. Imagine a researcher would like to generate his list of publications from the DLPB, Google Scholar, and CiteSeer. Suppose these scholarly libraries generate RDF of the results they provide, or a third service converts their results into RDF. Integrating and mashing up this content can be done easily using SPARQL, I shall illustrate this examples, in later posts.
It is important noting that Oracle 10g and 11g support RDF storage and querying. Oracle stores RDF triples in one simple table, in addition to some indices. Querying RDF in Oracle is done in a SPARQL style. As shown by Oracle, this implementation is scalable. For example, a query (with medium size complexity) over 80 Million RDF triples (5.2 GB) takes less than a second. It has been shown that retrieval cost per result row remains almost the same as the dataset size changes. I believe this simple and scalable support of the RDF technology (by Oracle) will accelerate the adoption of RDF as the web metadata format, and thus SPARQL as the web query language.
Posts
1 comment:
Mustafa, congratulations on the short article, very good.
I need your help so that I can continue my graduate work.
I need to create a basa Talis Platform to store an RDF base but I can not.
Could you give some tips?
My email is: peixecom23@hotmail.com
Thanks.
Perpetuo
Post a Comment