What Can You Model Using Metadata?
- Return and inputs of a "web service"
- Models of its functionality, not merely its I/O formats.
Descriptions like these will dramatically extend
UDDI and
ebXML specs.
- Individual field names of a data base
- Not just data types, but what the values represent.
Metadata enables many new services making information more flexible and portable.
- Similar tags in any XML document
- Not just document structure
and data formats (as in XML Schemas)
but arbitrarily complex models of what the tagged data signifies.
- Definitions for words and names
- Metadata is terrific for custom lexicons, a Lexikos specialty. They become
web-based, multilingual, expressive and mergeable at run time.
- Technical Hypertext Documents
- Specs for products and projects, using all the above, can now be
centrally managed as an XML resource for people or programs.
Such modeling powers - which are really AI techniques for web-based knowledge
representation - are a revolution in the making. They will impact your industry
this year, count on it! Call us to get their full benefits:
Example Applications
1. Metadata-based Transformers for Data Streams
Input data first loads into local Java objects reflecting a (meta)data
dictionary, reformatted under "rules" specific to each data source.
Driven by models of data types and topology, such conversion utilities
can handle restructuring; adapt readily to schema changes; flag data errors
per an ontology; and/or track the evolution of selected fields. They make
(e.g.) B2B data feeds or data migrations far simpler to set up and manage.
2. Report-Driven Intelligent Search Agents
These spider a specified mini-world of web pages as a private search agent, personalized to and by
their user(s) via metadata-modeled processes. Nightly or on demand, each (re)builds a private
report of web resources that is essentially a "smart" bookmark file tracking related interests.
To improve on (e.g.) current Google or UDDI queries, each agent can intelligently filter,
annotate and expand what they now return, and flag recent changes.
3. Software for Transcribing English to Metadata
Custom metadata lexicons describing your interests, plus Lexikos's own utilities for extracting
the information in written English, will let non-technical users create metadata in bulk by
supplying equivalent text and interactive transcription guidance as needed (not unlike a spelling
corrector). The XTM outputs can then be processed as simple XML data by other programs, who
can react to input text content (or filter it, translate it, route it, reply, etc.)
4. Learning and Reasoning Packages based on Metadata
In grad school I encountered the AI paradigms they use as "mini-theories" on how human
minds worked. They were good theories then, and for three decades since they have been
refined and extended by very clever people. Recasting them for J2EE means building Java
packages of such simulated mental powers, bundled so that clients can easily import or
download them to enhance intelligent agents (now emerging as "bots", "avatars", and accounts
in high end "content management" and "text mining" engines).
|
