RIF - Rules and the Semantic Web
From Semantic Web NYC
September 17, 2009
daylife 444 Broadway, 5th Floor New York, NY 10013
Who's coming?alex chan
Meetup with Chris Welty and discover Rules for the Semantic Web. The RIF rules standard is nearing completion as a formal W3C recommendation and various communities are beginning to implement tools that support it. In this talk Chris will introduce RIF, its design and goals, and discuss its potential uses on the semantic web.
Rule Interchange Format
Rule Interchange Format (RIF) is a proposed recommendation of the World Wide Web Consortium to provide a format for the interchange of rules in rule-based systems on the semantic web. The goal is to create an interchange format for different rule languages and inference engines.
RIF describes a number of dialects, initially including a Basic Logic Dialect (BLD) and Production Rule Dialect (PRD).
RIF can be viewed as a specification language for data. The presentation language uses the Prolog operator ':-', which places the conclusion of the rule first and the premises afterwards. As a result, RIF rules presented in this language are best read from the bottom up. RIF also incorporates a slot notation indicated with brackets to express assertions about the bindings that associate variables (on the left side of a binding) with values (on the right side of the binding).
The semantics of RIF are defined by interpreting each expression in the presentation language as a corresponding RDF expression. The semantics of an expression can be understood either by reading the presentation language expression or by reading the corresponding RDF. Usually, people will prefer to read the presentation language version.
The existence of variables and their bindings enables RIF to express constraints and relationships that a static schema expressed purely in RDF or RDFS could not express. RIF expressions can contain query-like effects, such as matching a value found in one construct with a value found in another construct. As with schema-based inference, computing the effect of RIF expressions involves deducing new data, but because of the variables, the effect of interpreting a set of RIF expressions with respect to a knowledge base is similar to running a program in a programming language on the data.
The BLD and PRD differ in the style of programming that they assume. The BLD assumes that the data will grow in a monotonic manner. The results that are inferred simply add to what was inferred earlier. Expressions in the PRD can be non-monotonic: evaluating the expressions can update or delete existing data.
Even in the absence of an engine that evaluates them, RIF expressions can be viewed as specifications. Because of their ability to name, place in correspondence, and specify transformation of various parts of a knowledge structure, RIF expressions can be used to describe transformations of data from one schema to another. These transformations can then be implemented in any programming language. If the chosen language is the language of a rules engine, the transformations can be accomplished either during import or by working within a knowledge base after import, depending on how the rules engine is integrated with the knowledge base.
RIF expressions can be used to express metadata assertions about documents. A convention is offered in the slides for attaching a metadata assertion as the immediate child of an XML element. This is read the same way as any standard XML serialization of RDF: the parent XML element is considered the subject of the relationship, the <meta> element names the relationship, and the contents of the <meta> element constitute the object. In this case, the object is a semantic annotation that makes an assertion about the subject.
Chris Welty is a Research Scientist at the IBM T.J. Watson Research Center in New York. Previously, he taught Computer Science at Vassar College, taught at and received his Ph.D. from Rensselaer Polytechnice Institute, and accumulated over 14 years of teaching experience before moving to industrial research. Chris' principal area of research is Knowledge Representation, specifically ontologies and the semantic web, and he spends most of his time applying this technology to Natural Language Question Answering as a member of the DeepQA/Watson team and, in the past, Software Engineering. Dr. Welty is a co-chair of the W3C Rules Interchange Format Working Group (RIF), serves on the steering committee of the Formal Ontology in Information Systems Conferences, is president of KR.ORG, on the editorial boards of AI Magazine, The Journal of Applied Ontology, and The Journal of Web Semantics, and was an editor in the W3C Web Ontology Working Group. While on sabbatical in 2000, he co-developed the OntoClean methodology with Nicola Guarino. Chris Welty's work on ontologies and ontology methodology has appeared in CACM, and numerous other publications.
The podcast is available to attendees on request.