Ontolog Panel Discussion: Rules in Semantic Web Applications - Thu 26-June-2008    (1HM3)

Conference Call Details    (1HM5)

Attendees    (1HMZ)

Agenda & Proceedings:    (1HN9)

Panel Discussion: "Rules in Semantic Web Applications"    (1HNA)

Topic: "Rules in Semantic Web Applications"    (1JU3)

The Panel:    (1JU4)

http://ontolog.cim3.net/file/work/Ontolog-Discussion/Rules-in-Semantic-Web-Application_20080626/LeoObrst_20080626.png . . . http://ontolog.cim3.net/file/work/Ontolog-Discussion/Rules-in-Semantic-Web-Application_20080626/MartinOConnor_20080626.png . . . http://ontolog.cim3.net/file/work/Ontolog-Discussion/Rules-in-Semantic-Web-Application_20080626/JosDeBruijn_20080626.png    (1JW1)

           [Dr. Leo Obrst]            [Mr. Martin O'Connor]         [Dr. Jos de Bruijn]    (1JW2)

Abstracts:    (1IL0)

The Rule Interchange Format (RIF) working group is about to publish the last call working draft of RIF BLD (Basic logic Dialect) [1], which is a Horn logic-based language for exchanging rules over the Web. In the Semantic Web context, it has been recognized that it must be possible to combine RIF rules with RDF data and RDFS and OWL ontologies, which led to the RIF RDF and OWL Compatibility specification [2]. RIF documents (i.e., rulesets) may import RDF graphs and OWL ontologies. The semantics of these imports is defined through combination of the RIF and RDF/OWL model theories. We give an overview of the design of RIF RDF and OWL compatibility, highlighting the decisions that led to the design, paying special attention to the way we dealt with the divide between the RDF and OWL DL semantics.    (1INA)

We then show how reasoning with combinations of RIF rules and RDF and a subset of OWL DL can be reduced to rule reasoning.    (1INB)

       [1] http://www.w3.org/TR/rif-bld/
       [2] http://www.w3.org/TR/rif-rdf-owl/    (1INC)

Jos de Bruijn received his PhD degree in Computer Science from the University of Innsbruck, Austria, in 2008 and his M.Sc. degree in Technical Informatics from the Delft, University of Technology, The Netherlands, in 2003.    (1JW3)

Since 2007 he is employed as assistant-professor at the Faculty of Computer Science, at the Free University of Bolzano, Italy. His research interests include Semantic Web (Services) languages, Logical languages, Logic Programming and Non-Monotonic Reasoning. He is (co-)author of 30+ peer-reviewed publications, including several books and journal publications.    (1JW4)

He has taught courses about the Semantic Web in the Master curricula of the University of Innsbruck and the Free University of Bolzano, Italy.    (1JW5)

Jos has been actively involved in the European projects COG, ASG, SEKT, DIP, and Knowledge Web. He is a member of the ESSI WSML working group and of the W3C RIF working group.    (1JW6)

Data integration is a central challenge of the Semantic Web. Applications built on the Semantic Web must typically deal with a variety of information formats. Ontology languages such as RDF and OWL provide a means of uniformly representing the semantic content of these formats. However, a variety of tools and techniques are required to physically transform the data in various formats to an integrated form. The transformations required range from low level structural or syntactic mappings to mappings that require extensive domain knowledge. We show how rules can provide a unifying knowledge-based approach to building some of these mapping technologies. In particular, we show how the Semantic Web Rule Language (SWRL) can be used to tackle the complex structural and semantic transformations needed to reconcile common Web information formats. We describe a number of open-source tools we have developed to allow users to perform these mappings.    (1IL5)

Martin O'Connor is a currently member of the Research Staff at the Center for Biomedical Informatics Research at the Stanford Medical Schopol (Stanford-BMIR). Before joining Stanford-BMIR, he worked for several years at IBM's T.J. Watson Research Center in Yorktown Heights, N.Y. At Stanford he was involved in the development of the EON system, a component-based architecture for developing decision support systems for guideline based care. I also developed the Chronus temporal query system which was used to perform temporal queries on biomedical data. He is currently involved in the development of BioSTORM, a research system to evaluate intelligent systems for epidemic detection and characterization. In parallel with the BioSTORM efforts he has been investigating the use of rules and query languages on the Semantic Web. Mr. O'Connor currently developing the SWRLTab, an open source Protege-OWL-based development environment for the Semantic Web rule language SWRL.    (1JW7)

Ontologies enable explicit expression of community concepts and support Machine-to-Machine (M2M) interactions at the semantic level. Ontologies expressed in a standard language such as OWL and exposed on a network offer the potential for greater interoperability solutions since ontologies are semantically rich, computer interpretable and inherently extensible. In this talk, we describe how we applied ontologies and rules for rapid enterprise integration of heterogeneous data sources and aggregation of events for situational awareness in a pilot focused on military convoy movement in a theater operation. In particular, we describe our transformation of OWL ontologies + SWRL rules into the logic programming paradigm, and the resulting efficient runtime automated reasoning architecture that utilized those ontologies, knowledge bases, and rules in a service-oriented framework that also used Google Earth for visualization.    (1IL2)

Leo Obrst is a principal artificial intelligence scientist in the Information Discovery and Understanding department at MITRE's Command and Control Center, where he leads the Information Semantics Group (semantics, ontological engineering, knowledge representation and management), and has been involved in projects on Semantic Web rule/ontology interaction, context-based semantic interoperability, ontology-based knowledge management, conceptual search and information retrieval, metadata and taxonomy/thesaurus construction for community knowledge sharing, intelligent agent technology, semantic support for natural language processing, and ontology-based modeling of complex decision-making. He is also currently involved in many US federal government efforts to establish Communities of Interest (COI) vocabularies/models for information sharing, including the development of universal and common core vocabularies/models which span those COIs.    (1JW8)

Leo's PhD is in theoretical linguistics with a concentration in formal semantics from the University of Texas-Austin. He has worked over 24 years in computational linguistics, knowledge representation, and in the past 14 years in ontological engineering and more recently in Semantic Web technologies. Besides MITRE, Leo has worked for VerticalNet, Boeing, Software Development Group (a partner), Intelligent Business Systems, the Microelectronics Computer Corporation (MCC), Texas Instruments, SoftCraft, and Ohio Edison. Leo is also one of the three co-conveners of ONTOLOG, a virtual community which they co-founded in 2002.    (1JW9)

Resources    (1JU5)

Questions, Answers & Discourse:    (1JUC)

Questions and Discussion captured from the chat session:    (1JUT)

 PeterYim: Welcome to the Ontolog Panel Discussion: Rules in Semantic Web Applications - Thu 26-June-2008    (1JUU)
    * Panel Session Topic: Rules in Semantic Web Applications
    * Session Chairs: Dr. Leo Obrst (MITRE; Ontolog)
    * Panelists / Title of Presentations:
          o Dr. Jos de Bruijn (Free University of Bozen-Bolzano, Italy) - "RIF: A Standard Rules Language 
                                                                           for the Semantic Web"
          o Mr. Martin O'Connor (Stanford-BMIR) - "Knowledge Integration with SWRL"    (1JX1)
          o Dr. Leo Obrst (MITRE) - "Semantic Web Ontologies and Rules With Efficient Runtime Reasoning 
                                     Using Logic Programming"    (1JX2)
 TomBrunner: Where does F-Logic fit into ontology building?    (1JX3)
 ArturoSanchez: Are these mappings and combinations automatically derived or people have to build them 
                on a case-by-case basis?    (1JX4)
 JosDeBruijn: Note that RIF-OWL DL combinations are essentially a superset of SWRL.  The semantics of 
              these combinations is in the spirit of SWRL.  You can use exclusively OWL classes and 
              properties, but you can also use other predicates.    (1JX5)
 PatCassidy: Does SWRL/SQWRL support OWL-full?    (1JX6)
 JosDeBruijn: PatCassidy: No; SWRL is based on the OWL DL semantic world, which is incompatible with 
              the RDF semantic world (OWL full is based on the latter)    (1JX7)
 JosDeBruijn: Pointer for Martin: Parsia & Co. created a SPARQL-like language for querying OWL DL; 
              I expect this can be readily adapted to SWRL: http://clarkparsia.com/files/pdf/sparqldl.pdf    (1JX8)
 MartinOConnor: SWRL/SQWRL does not directly support OWL-Full. However, there is a TBox built-in library 
                that can be used (http://protege.cim3.net/cgi-bin/wiki.pl?SWRLTBoxBuiltIns) that provided 
                effective support, though it would not be safe to use these built-ins to reach conclusions; 
                we use them a lot in queries.    (1JX9)
 Xenia Fiorentini: when will the SWRL Tab be available for Protege 4?    (1JXA)
 Xenia Fiorentini: is there a tool/editor you would suggest me to use to work with both OWL 1.1 and SWRL?    (1JXB)
 MartinOConnor: Protege 4 supports OWL 1.1 and has a very limited SWRL editor. Pellet can be used 
                in Protege 4 to execute the rules. Pellet now has pretty robust SWRL support and 
                I think understands OWL 1.1. I hope to port all SWRLTab functionality to Protege 4 
                by the end of the year.    (1JXC)
 Xenia Fiorentini: thank you Martin    (1JXD)
 KenBaclawski: Jos de Bruijn: I have a question related to Arturo's.  How general is this notion of profile? 
               Will there be a small set of standard profiles, or could one define other profiles?    (1JXE)
 JosDeBruijn: Ken: there is a set of defined profiles, but this set 
              is extensible: http://www.w3.org/2005/rules/wiki/SWC#Specific_Profiles
              ... (there is a comment about new profiles just above the sections)    (1JXF)
 KenBaclawski: Martin O'Connor: I did some work on integrating highly dynamic database queries with 
               rules in wireless communication.  We used prolog as the rule engine and the "database" 
               was a running program which was queried using programming language reflection.  See 
               D. Brady, K. Baclawski, M. Kokar and J. Wang. Achieving Self-Awareness of SDR Nodes through 
               Ontology-Based Reasoning and Reflection. 
               ... In Proceedings from the Software Defined Radio Technical Conference - SDR'04 (2004).    (1JXG)
 MartinOConnor: Ken - thanks. I will take a look at that paper.    (1JXH)
 JosDeBruijn: RDFS cannot be translated to SQL; ground entailment in RDFS is Ptime-hard, whereas 
              with SQL you can only compute LogSpace queries    (1JXI)

Audio Recording of this Session    (1JUW)