Outline

• An introduction to the Semantic Web
• Why Biblical scholars should care
• New Testament Names (NTN): a semantic knowledgebase
• Going further

Semantic Web in the News

Key Concepts for the Semantic Web

• Principal goal: a web of data that allows machines to "connect the dots" and share across application (and organizational) boundaries
  • Other uses: browsing, visualization, inference, statistical analysis ...
• Everything has a URI
• The RDF triple is a primary data element
• RDF/XML standardizes syntax: OWL standardizes semantics to enable data interoperability
  • Application-specific semantics are not enough

Web Ontology Language (OWL)

• OWL has been an official W3C recommendation since 2004. It starts from XML and RDF and adds
  • Ontology specification
  • Property characteristics (transitivity, symmetry, functionality)
  • Property restrictions (cardinality, value restrictions)
• An ontology is a representation of a conceptualization (Gruber 1993)
• See the OWL Guide at http://www.w3.org/TR/owl-guide
• Beyond OWL: reasoning, logic, ...

The Semantic Web Process

• Model your data (conceptualization)
• Develop an OWL ontology that reflects that model (representation), reusing existing ontologies where they fit
• Populate instance data (preferably by mapping existing data sources)
• Connect to other data
• Deploy "added value" tools like visualization, inference, and validation

Semantic Web and Text Markup

• Ontologies describe what is: real things, not documents
• XML for text markup, OWL for "knowledge markup"
  • XML can still connect to the Semantic Web through rdf:ID
  • Topic Maps are perhaps somewhere between XML and OWL
• Folksonomic tags (del.icio.us, Technorati) are a different bottom-up alternative

Semantic Web Tools

• Editors: Protege, SemanticWorks, others
• Triple Stores: Sesame, Oracle Spatial 10g, Joseki, Tucana
• Programming libraries for Java (Jena), Perl, Python, etc.
• Longwell is useful for browsing RDF
• SemWeb search through Swoogle: currently > 300M triples
• Also reasoners, validators ...
• W3C has a much more comprehensive list at esw.w3.org/topic/SemanticWebTools

Example: Geodata Integration

• Google Earth data for Bible Places (John Philip and others) provides latitude, longitude, and altitude in Keyhole Markup Language (KML)

<Placemark>
<name>Antipatris</name> …
<LookAt>
	<longitude>34.954…</longitude>
	<latitude>32.1068…</latitude>

Google Earth Visualization

• 
• "... the next day we touched at Samos; and the day after that we went to Miletus. For Paul had decided to sail past Ephesus ..." (Ac 20:15-16, ESV)

The Integration Problem

• Variation in the <name> element
  • “Ashtaroth, Astaroth”
  • “Akkad (Accad)”
  • “Acco / Ptolemais”
  • “Antioch” (Syria), “Antioch” (Pisidia)
  • “Ar / Ar of Moab”
  • “Bethany”, “Bethany Beyond the Jordan”
• Type variation: “Sea of Galilee”, “Mount of Olives”, "Solomon's Porch"
• How can we form semantic connections to this data?

A Semantic Web Solution

• Give resources a URI like

  http://semanticbible.org/ns/2006/NTNames.owl#Antioch

• Every resource has a type:

  <http://semanticbible.org/ns/2006/NTNames#Antioch>
  
  <http://www.w3.org/1999/02/22-rdf-syntax-ns#type>
  
  <http://semanticbible.org/ns/2006/NTNames#City> .	

• Use rdfs:label to capture alternate names and spellings
• Use owl:sameAs to connect resources across ontologies

New Testament Names: a Semantic Knowledgebase

• An instance for each unique entity named in the New Testament
  • Provides unambiguous identifiers
• Provides an initial reference ontology with instances
  • 48 classes, 38 properties
  • 647 instances, 4200 RDF triples
• The content is in the identifiers and their relationships, rather than prose
• (Link to overview description)

New Testament Names (2)

• Predominant classes:
  • Man (330 instances)
  • Woman (46 instances)
  • City (90 instances)
• NTN is the first and (so far) only publically available Bible data in OWL
  • First version released 2004-04
• The ontology and instance data are freely available
  • Ontology: http://www.semanticbible.org/2006/11/NTNames.owl
  • Instances: http://www.semanticbible.org/2006/11/NTN-individuals.owl (the instance data).

New Testament Names: Classes

• Classes are defined consistently with existing ontologies where possible
  • A comment on ontology reuse
• OWL supports multiple inheritance

New Testament Names: Properties

• Humans have attributes like description, occupation, name_en, name_gk
• Properties between Humans and
  • Humans:
    • family relations (parent/child, spouse, sibling)
    • inter-personal relations: collaborates with, antagonist of, enemy of, knows
  • Locations: native place, resident place, visited place
• Religious belief, ethnicity, group membership
• Locations have latitude/longitude, spatial inclusion

New Testament Names: Instances

• URIs disambiguate Alexander (5), Joseph (4), James (3), Mary (6), etc.
• Principles:
  • Limit to what's explicit in the text
  • Conservative grouping
• Browser demonstration using Longwell: Cities in Macedonia

An Integration Example: Connecting the Composite Gospel and NTN

• Pericope.121: Herod kills John the Baptist
• Add properties for textual mentions of Herod, Herodias, John, Philip (and others who aren't named) using NTN identifiers
• Now NTN data is available to enrich a display or other application

Extending NTN

• Other relevant entities that are either
  • mentioned in the text without names
  • relevant to the context of New Testament study
• Add Bible references as properties; annotate textual mentions with NTN URIs
• Incorporate other reference sources (e.g. Wikipedia's category for New Testament people)
• Incorporate events and time (support timeline visualization)
• Connect to specific vocabularies (theological terms, historical/cultural background)

Other OWL Applications for Bible scholars

• OSIS is a complementary step toward standardizing text markup
• Modeling the life of Jesus:
  • Composite Gospel (RDF) describes relationships between textual units of the Gospels
  • Events of His life
  • Concepts that He taught
• Authorship information (the Biblical canon, Ante-Nicene Fathers, etc.)
• Lexical semantics (could we integrate Wordnet and Louw-Nida?)
• Topical markup

Some Possible Objections (and Responses)

• Requires black-and-white judgments about inherently "gray" matters
  • The ontologist is responsible for the model
• Requires too much effort and technical savvy
  • Tools are improving
  • Opportunistic beginnings provide important foundations
• Limited present value
  • Standards are always forward-looking
  • Agreement is easier before competing standards become entrenched

An Invitation to Broader Participation

• The primary data are open to enrichment and extension
• Metcalfe's Law applies here too
• Biblical scholarship would benefit from greater web-enablement: this requires semantic standardization

References

1. "Why use OWL?" Presentation by Roger L. Costello and David B. Jacobs (MITRE).
2. "Semantic Web: Building on what exists", by Tim Berners-Lee (W3C), http://www.w3.org/2006/Talks/0404-mit-tbl
3. "Semantic Web concepts", by Tim Berners-Lee (W3C), http://www.w3.org/2005/Talks/0517-boit-tbl
4. "Questions (and Answers) on the Semantic Web", by Ivan Herman (W3C), http://www.w3.org/2006/Talks/0927-Berlin-IH
5. "A translation approach to portable ontologies", by T. R. Gruber, Knowledge Acquisition, 5(2):199-200, 1993.