90 likes | 222 Views
OWLIM - OWL DLP support within Sesame Damyan Ognyanov Ontotext Lab, Sirma AI. Introduction. OWLIM is an OWL DLP In-Memory SAIL for Sesame 1.1. SAIL = storage and inference layer OWLIM supports partial, forward chaining based, reasoning over OWL DLP . It is open-source
E N D
OWLIM- OWL DLP support within SesameDamyan OgnyanovOntotext Lab, Sirma AI
Introduction OWLIM is an OWL DLP In-Memory SAIL for Sesame 1.1. • SAIL = storage and inference layer • OWLIM supports partial, forward chaining based, reasoning over OWL DLP. • It is open-source • It includes a custom RMI factory that enables an application to get direct access to the sail-stack of a repository through RMI. OWLIM is: • developed by Ontotext • used in SEKT, through the KIM semantic annotation platform • A “reference reasoner” for the PROTON upper-level ontology OWLIM, OM Tools Fair, Grenoble
Conceptual Grounds • Forward-chaining of entilement rules like those used in the RDFS model-theoretic semantics: <X type Y> && <Y subClassOf Z> => <X type Z> • OWL primitives are modeled as: <X Prop Y> && <Y Prop Z> && <Prop rdf:type owl:TransitiveProperty> => <X Prop Z> OWLIM, OM Tools Fair, Grenoble
Functionality • The functionality of Sesame: • Parsing, serialization of RDF (XML, NTriples, N3) • Storage and retrieval • Reasoning • Query Answering (RQL, RDQL, SeRQL) • OWL RDF schema supported, including: SymmetricProperty, TransitiveProperty, inverseOf, equivalentClass, equivalentProperty, sameAs, FunctionalProperty, InverseFunctionalProperty OWLIM, OM Tools Fair, Grenoble
Persistence Strategy • OWLIM “reasons” in-memory, but has a comprehensive persistence and backup strategy • Persistence based on N-Triple files: • Number of files can be given as “pre-loaded” • Only one of them can be updated (the “main trunk”) • Special attention paid to assure that no loss of information or inconsistency can be caused by an unpredicted interruption • The strategy for synchronization of the in-memory representation with the persistent files is configurable • The persistence strategy has been extensively tested • It has been used in KIM for a couple of years • It is more reliable than many of the DB-based repositories OWLIM, OM Tools Fair, Grenoble
Client3 Client1 Client2 H T T P R M I SOAP The Architecture: SESAME … HTTP Handler SOAP Handler Remote Access Repository Services Modules Admin Module Query Module Export Module SAILs Storage And Inference Layers (SAIL) DB, Files, … OWLIM, OM Tools Fair, Grenoble
Performance and Scalability • One of our load tests: • Take KIMO (the PROTON Predecessor) ontology • About 300 classes and 100 properties; • The initial KB contained 700k entity descriptions - the KIM World KB was extended with the entities and properties, extracted from the top news for the period 2002-2004. • The test was performing transactions of an addition of 1000 synthetically generated entity descriptions. • Average number of statements, per transaction: 12k. • Average commit time: 9 sec. • Running on 2xOpteron (1.4Mhz) with 6GB of RAM worth $2000 • JDK 1.5 64-bit edition OWLIM, OM Tools Fair, Grenoble
Performance and Scalability (II) • The results: • Hosted 1.2M entity descriptions: • about 15M explicit statements; • above 30M statements after forward chaining. • 1300 statements/second upload and reasoning • Linear growth of time and space required • 170 bytes/statement in-memory • N3 storage up to 2.2GB, 147 bytes/statement OWLIM, OM Tools Fair, Grenoble
Current Status and References for OWLIM • http://www.openrdf.org– the home of Sesame • http://www.ontotext.com/downloads/index.html#sesame-sail • The distribution contains: • Sources and Binaries • System Documentation • Short term plans: • Development related to ORDI – support for WSML Core • Provide full support for OWL DLP OWLIM, OM Tools Fair, Grenoble