SPOC-Web Icon, semantic Knowledge Management

Semantics in Depth

This Page goes even deeper into the semantic Rabbit Hole to explain some of the Decisions made in the Spoc-Web Implementation. We try to avoid confusing Details and to convey the broad and important Ideas without sacrificing too much Accuracy. We hope not to violate Correctness though. Please feed back any Errors You find in these Documents.

First Order Logic, Languages and Inference

One of the Goals for the semantic Web is for the Computer to "understand" the Data in the World Wide Web. Among others this means, the Computer can

  • collect, filter and present semantic Data
  • search for related Information on the Internet
  • allows to formulate complex Queries involving multiple free Variables (see SPARQL)
  • understand the Context of natural Language Queries
  • prove or falsify Propositions
  • deduce new Information from Data and Metadata using the Rules of Logic

The Calculus for these Operations is called First-Order- or Predicate-Logic. The Algorithms for this are an intense Research Topic since the first Artificial Intelligence Wave in the early 1980s. Unfortunately their Runtime Performance is often very low and there is a Risk of non terminating Processes. Additionally the näive Handling of Rules and Proposition can allow to construct inconsistent or contradictionary Systems.

Since then Research has identified several Levels of Languages together with their associated Inference Engines/ Algorithms with reasonable Performance. Also, as Experience shows, the predicted Worst Case Runtimes are rarely encountered. Thus the current Reasoning Engines become useful for a growing Number of Applications.

The Distinction between these Languages usually too detailed for the Purposes of Spoc-Web, therefore several closely overlapping Concepts between these Languages, together with their URIs have been merged into unique Conceps in Spoc-Web.

Languages

The Logic Languages in this Sections are ordered by "Inferential and Expressive Power". Usually this Power also increases the Complexity and decreases Performance.

  • RDF is a simple Language to write Meta-Data about URIs
  • RDFS defines Classes, but in such an unprecise way, that Inference can lead to wrong Results
  • OWL Ontology Inference Language (the Acronym is dedicated to the Owl Character from A.A.Milnes Book Winnie the Poo)
  • OWL 2 extends and details the first OWL Standard into 2 Flavors with 3 Profiles each to balance Expressivity with Performance:
  • OWL 2 Flavors: DL and Full
    • OWL DL (Direct Logic) distinguishes the Instance and the Class Role of a Class Resource/URI. This avoids Confusion, inifinite Recursion and increases Performance.
    • OWL Full allows the dual Use of the Class and Instance Roles and is therefore undecidable and computationally expensive.
  • OWL 2 Profiles: EL, RL and QL
    • EL = Expression Logic: allows only existential Quantifications, no universal (All) and no Negation/Inverse
    • QL = Query Language: this Language allows to translate Logic Queries into relational Queries
    • RL = Rules Language: reasoning in this profile is already implemented in many standard Rule Languages

Predefined IDs for core Entities

Spoc-Web is designed to share Data between Databases. This requires a common Language that is established by fundamental RDF Classes and Relations defined by the W3C in the Namespaces rdf, rdfs and owl.

While the W3C uses URIs to denote Things, Spoc-Web uses integer Numbers for Speed and Compactness. To facilitate Integration, most of the fundamental Classes have fixed IDs below 1000. This allows Applications to hard-code them into their Code. These Definitions are included in the initial Database You downloaded together with the Spoc-Web Application.

Among these are:

 

Core RDF Namespaces

These are some of the important Namespaces defined in Spoc-Web. You can also directly inspect them in Spoc-Web itself, e.g. by entering their ID into the Search Dialog.

ID URI Prefix Description
330 rdf http://www.w3.org/1999/02/22-rdf-syntax-ns# RDF Core Definitions like Resource, Class
332 xsd http://www.w3.org/2001/XMLSchema# XML Value Types like string, float, int etc.
331 rdfs http://www.w3.org/2000/01/rdf-schema# RDF-Schema Definitions
333 owl http://www.w3.org/2002/07/owl# OWL-Definitions
334 foaf http://xmlns.com/foaf/0.1/ Friend of a Friend Classes, an early and widespread RDF Language used to describe Relations among People
335 dc http://purl.org/dc/elements/1.1/ Dublin Core Classes, defined by Librarians to describe Books and Documents in the World wide Web
337 dbp http://dbpedia.org/ DBPedia Core Namespace, heavily used in Wikipedia
338 dbpprop http://dbpedia.org/property/ DBPedia Property Definitions
339 dbpedia http://dbpedia.org/resource/ DBPedia Individuals/Things
362 yago http://dbpedia.org/class/yago/ A unified Ontology, merged from WordNet and DbPedia
364 schema http://schema.org/

Namespace of Schema.org, a Consortium created by Bing (Microsoft), Google and Yahoo! to to “create and support a common set of schemas for structured data markup on web pages.”

363 umbel http://umbel.org/umbel/rc/

short for Upper Mapping and Binding Exchange Layer, is a strict subset of OpenCyc, providing the Cyc data in an RDF ontology. Because it is subset of OpenCyc concepts and relationships, mappings to UMBEL can take advantage of the reasoning engines within Cyc.

365 dbpedia-owl http://dbpedia.org/ontology/ The DBpedia Ontology is a shallow, cross-domain ontology, which has been manually created based on the most commonly used infoboxes within Wikipedia.
505 swap http://www.w3.org/2000/10/swap/ "Semantic Web Application Platform", some of the first Definitions for the Semantic Web by the W3C
367 cyc http://sw.cyc.com/concept/ Cycorp "Cyc" Namespace, a first Generation AI (Artificial Intelligence) Project mostly financed by the USA DARPA (Defense advanced Research Project Agency).

 

Core RDF Classes

ID Prefix Suffix Description
 323 rdfs Resource These two Classes have been merged into a single Spoc-Web Class called "Resource~Things", because their Distinction is too narrow for general Purposes.
owl Thing