[announcement] BigOWLIM Reasons over 1 Billion statements of OWL Passing LUBM(8000,0)

(*** We apologize if you receive multiple copies of this announcement ***)

BigOWLIM successfully passed the threshold of 1 billion (10^9) statements of 
OWL/RDF - it loaded
an 8000-university dataset of the LUBM benchmark 
(http://swat.cse.lehigh.edu/projects/lubm/) and
answered the evaluation queries correctly.

Evaluation setup and statistics:
 - Hardware: 2 x Opteron 270, 16GB  of RAM, RAID 10; assembly cost < 5000 
EURO
 - OS: Suse 10.0 Linux, x86_64, Kernel 2.6.13-15-smp; 64-bit JDK 
1.5 -Xmx12000m
 - Loading, inference, and storage took 69 hours and 51 min
 - LUBM(8000,0) contains 1.06 billions of explicit statements
  -- The "inferred closure" contains about 786M statements
  -- BigOWLIM had to  manage over 1.85 billions of statements in total
 - 92GB RDF/XML files; 95 GB binary storage files
 - Average Speed: 4 538 statements/sec.

OWLIM (http://www.ontotext.com/owlim/) is a high-performance semantic 
repository,
packaged as a Storage and Inference Layer (SAIL) for the Sesame RDF database 
(v.1.2.1-1.2.4).
OWLIM uses TRREE (Triple Reasoning and Rule Entailment Engine) to perform 
RDFS, OWL DLP,
and OWL Horst/Tiny reasoning. The most expressive language supported is a 
combination of unconstrained
RDFS and limited OWL Lite.

BigOWLIM (http://www.ontotext.com/owlim/big/) is a new and even more 
scalable version of the
OWLIM semantic repository. A pre-release of BigOWLIM (ver. 0.9-beta) is 
already available.
The "standard" OWLIM version, which performs reasoning and query evaluation 
in-memory, is referred
now as SwiftOWLIM. The major distinctive features of BigOWLIM are:
 - The storage, inference and query evaluation are performed directly 
against binary files; it does not need to
maintain all the contents of the repository in the main memory; this allows 
instant startup and initialization of
large repositories;
 - BigOWLIM performs database-like query optimizations; re-ordering of the 
constraints in the query has no
 impact on the execution time;
 - BigOWLIM features special handling of owl:sameAs, so that processing of 
large "classes" of equivalent
resources, does not cause excessive generation of inferred statements;
 - It has much lower memory requirements: LUBM(50,0) can be handled in 26 
min on a desktop machine,
giving Java (-Xmx) 192MB of RAM; LUBM(1000,0), which includes over 130 M 
statements, is handled
in 11h on the same machine, given 1.6GB of RAM.

We also take the opportunity to announce the release of ver. 2.8.3 of 
SwiftOWLIM (the in-memory version),
which features several improvements and better configurability. The major 
changes in version 2.8.3,
with respect to 2.8.2, are:
 - Improved concurrency: several fixes took place to allow swift handling of 
hundreds of simultaneous users;
 - Stack-safe mode: a new parameter (stackSafe) allows switching the engine 
in a slower mode, which prevents
stack overflows that could happen for some datasets and ontologies in the 
standard mode;
 - eLUBM: in stack-safe mode, SwiftOWLIM passes the Lite-1, Lite-5, DL-1, 
and DL-5 tests of the
eLUBM benchmark on a desktop machine given 512MB. eLUBM is an extended 
version of the LUBM test:
"Towards a Complete OWL Ontology Benchmark", [Li Ma et al], ESWC2006;

OWLIM will be presented today (25th of May, 14:00 GMT) in the Developers 
Track of WWW2006,
http://www2006.org/programme/ .


Damyan Ognyanoff,
Ontotext Lab, Sirma Group Corp.
http://www.ontotext.com

Received on Thursday, 25 May 2006 10:59:32 UTC