Working with the Ontos Architecture

1
Working with the Ontos Architecture

• Where were going
• What you should know

2
Overview

• Current and future architecture
• Backwards compatibility
• Working with OSMX documents
• CVS techniques
• Java tips, resources, best practices

3
Current Architecture

• Ontologies OSM-L, osm.dtd
• Ontology Editor
• Data Frames rev. 2 support
• Runs as applet or standalone app
• Automatic layout algorithm
• Reads osm.dtd-based docs
• Writes to various formats (OSM-L, osm.dtd,
  PostScript)
• Ontos
• Handles XML and RDF as well as old files
• Produces SQL insert statements
• OSM-L parser, record boundary detector, etc.

4
Future Architecture

• Ontologies OSMX (XML Schema)
• Store data directly with ontology
• Ontology Editor
• Support new data frames ideas
• Multiple phrases with confidence values
• Specification of data types and units of measure
• Methods
• Import old files write OSMX
• Data frame library management
• Ontos
• Gen/spec working with data frames
• Work with new extraction framework

5
Backwards Compatibility

• We dont want to lose or have to re-create old
  ontologies
• Choice support old or convert them?
• Support could be difficult over time
• Conversion auto, manual, hybrid?
• XML Transform (XSLT) will help with conversion
  but not all the way
• Manual tedious, one-time effort
• How many ontologies are we talking about?

6
Working with OSMX

• Reading and writing OSMX
• Use JAXB-generated classes

// First get a JAXBContext instance from the
factory method // Pass in the package containing
the binding classes JAXBContext ctxt
JAXBContext.newInstance("edu.byu.deg.osmx.binding"
) // We now wish to get an in-memory object
tree from the source file Unmarshaller u
ctxt.createUnmarshaller() OSM osm (OSM)
u.unmarshal(new java.io.File("some-file.xml")) /
/ From here the OSM object may be used and
modified however you please. // Now we wish to
write the in-memory objects to a destination
file Marshaller m ctxt.createMarshaller() m.mar
shal(osm, new java.io.FileWriter(new
java.io.File("some-result.xml")))
7
OSMX Structure
ObjectSet
NameList
Name
DataFrame
InternalRepresentation
ValuePhraseList
ValuePhrase
KeywordPhraseList
KeywordPhrase
MethodList
OSM
8
Sample OSMX Document
ltOSM xmlnsxsi'http//www.w3.org/2001/XMLSchema-i
nstance' xsischemaLocation'http//www.deg.byu.
edu/xml/osmx.xsd
http//www.deg.byu.edu/xml/osmx.xsd'
xmlns'http//www.deg.byu.edu/xml/osmx.xsd'
order"1"gt ltMacro label"SomeMacro"gt0-9lt/Macr
ogt ltNotegtltText order"2" content"Some
note."gtlt/Textgtlt/Notegt lt/OSMgt
9
Available Projects

• Data frame library management tool
• Data instance validator
• Convert and validate old ontologies
• Automatic precision/recall calculator (store and
  reuse hand-tagged data)
• OSM-L to OSMX converter (use existing C parser to
  emit XML?)
• Data instance to SQL insert converter
• OSMX to OSM-L stylesheet (XSLT)

10
CVS Techniques

• Tagging Store a milestone
• Branching Parallel development
• Steps
• Create tag/branch on selection or entire module
• Update from repository, specifying tag/branch
• Tag/branch sticks to your local working files
  when you commit changes
• Merge branches when done
• Switch to trunk or other branch using update

11
Java Best Practices

• Coding standard
• Follow Suns example
• See Suns The Java Tutorial for samples
• Use javadoc comments
• Avoid hard-coded constants
• Have editor expand tabs to spaces
• Indent two spaces
• Use white space thoughtfully

12
Using Java

• Stay up-to-date but avoid bleeding edge where
  possible
• Java 1.2 Swing (instead of AWT)
• Java 1.3 Collections (instead of Vector,
  Hashtable, Enumeration)
• Java 1.4 javax.xml, java.nio, asserts, logging,
  preferences,
• Java 1.5 (coming) generic types, etc.
• Check within code for support of new features so
  older VMs fail gracefully

13
Resources

• java.sun.com
• Tutorials
• API documentation
• developer.java.sun.com
• TechTips e-mail newsletter
• Free tools
• Suns stuff, of course
• JBuilder, Eclipse

14
Additional Resources

• xml.apache.org (xerces, xalan)
• w3.org (XML specs and resources)
• xml.com (XML tutorials)
• netbeans.org (good free Java IDE)
• CVS repository
• DEG website