Intro, Orri Erling

Hi

I am Orri Erling, program manager for OpenLink Software's Virtuoso product.

My earliest background is in Lisp compilers, with some interests touching on
inference and knowledge representation.  Subsequently, I have worked on all
aspects of databases, from storage engine to SQL optimization, transactions
etc.  I am the  principal author of Virtuoso's DBMS core functionality and
SQL compiler.

My most recent activity represents  a revisiting of database basics, with a
rewrite of all engine thread synchronization structures for better SMP
performance.  This is not specific to RDF but RDF processing benefits
directly.  Immediate future plans include some SQL extensions for better
processing of SPARQL queries accessing relational data mapped to RDF and
further RDF oriented data type support in the Virtuoso engine.

RDF and SPARQL, representing the new frontier in databases and information
integration, as well as the coming together of inference and databases,
directly corresponds to my interests of all time.  Hence I am looking
forward to participating in this process.

Since OpenLink is a provider of data access technologies, from data access
drivers to Virtuoso's federated database capabilities, OpenLink sees RDF is
the completing piece of the data integration story.  Hence our interest in
and extensive work in mapping legacy data to RDF.

We see RDF and SPARQL as potentially becoming the lingua franca of
heterogenous data access, as well in the enterprise as on the web.

The web will continue with tagging and full text search.  The enterprise
will continue with the current range of EDI, EAI, data warehousing etc.
These are distinct specialty domains, often dealing with relatively separate
data silos.  RDF and SPARQL promise a new range of flexibility in publishing
and joining this data in unforeseen and innovative  new ways.

In order to fully fit this role, SPARQL needs to grow slightly, extending
itself to the web with the inclusion of text search and to the enterprise
with aggregation and grouping.  Since hardcore business intelligence
applications will not overnight migrate to RDF, the aggregation and analysis
does not have to be very sophisticated, maybe at the level of SQL 92 with no
OLAP extensions.  Even for supporting faceted search, which is becoming a
prevalent UI paradign specially for the semantic web, a count operator is
needed.

Another need that we perceive is in the domain of SPARQL end point
self-description, specifically for discovery, supporting federated queries
and describing language extensions.

We understand that the DAWG is not presently occupied with these matters
but we hope to see them on the agenda in the future.

My colleague Ivan Mikhailov, here present, is the author of the SPARQL to
SQL front end and relational to RDF mapping in Virtuoso.  hence he is best
qualified to be OpenLink's primary representative  on DAWG.

I  write commentary  on RDF and database topics from time to time on the
Virtuoso product blog:

http://virtuoso.openlinksw.com/blog

Some of our RDF related work is documented on papers available from:

http://virtuoso.openlinksw.com/wiki/main/Main/VOSArticles



Regards
Orri Erling

Received on Monday, 22 January 2007 09:54:10 UTC