Applying Semantic technology to KM

Wednesday, July 18th, 2007
Chris Bunk, Knowledge Systems Engineer for of Ares Corporation, up next; he's presenting on applying semantic technology to the practice of knowledge management.

Disclaimer: Semantic technology is a tool, not a solution. great slide

Think back to the day
- When data was typed and stored in file cabinets
- Difficult to know what you had or to find things
- even harder to know what someone else might have in their file cabinet

Fast moving presentation. I'll listen to audio and try to fill in later.

Disparate systems
Challenges
- Collecting data required visits to numerous systems
- Data in one system may not use same conventions as other systems
- No way to do complex queries across entire system without a high level of manual effort
- Searching through records containing large amounts of unstructured text requires knowing the right keywords to find what you are looking for.

Design system
Challenges
- Reusing previous designs is only possible if the engineer knows what [past] project and part to look at
- Searching for past problem reports often requires extensive searching
lessons learned that could be applied to design process are typically captured with minimal meta data.

Duplicative systems
Challenges
- manually converting all the data from one system to another required tremendous effort
- each system uses its own codes, terminology, and data standards
- Data owners can be reluctant to show their "skeletons in the closet."

Decisions Support System
- Analysis across disparate system requires considerable effort.

These problems require a new way of thinking...

Think of what information is in your organization, how does it interrelate, what business function is it tied to and how to use it.

Map out all of the information in your organization
Determine what you want to be able to do with this information
Determine who will be able to see what data and who has the ability to add, edit, and delete the data.
Only once you have defined your ontology and use cases should you even begin to think about your IT infrastructure.,

Need a common ontology - information about how to describe information (See slide/WikiPedia for example.)
Eric: see http://nasataxonomy.jpl/.nasa.gov/2.0/wbs for an example.

What tools and technologies do I need to implement semantic features?
- TRANSPORT to move the data (TCP/IP)
- REPRESENT information and relationships (XML, common, also RDF, RDFS, OWL, and SWIRL)
- A place to STORE this information
- A way to QUERY  this information (SPAR QL - the SQL of the Semantic web) Look at opensource engines e.g. Pellet for semantic tools)
 - There need to be TOOLS to develop the ontologies and manipulate the Semantic data.. analyze, search, process

 See quote from Tim Berners-Lee on Semantic web (Slide 12)

Software tools mentioned:
Insight  (might be Inxight sp?)
Clearforest
Lexolytics
Autonomy
Goldfire
Downside on last two - they don;t do a good job of letting you use the relationships for other purposes. They are mature tools, so weigh the issues.

Thesis; Maintain control of your knowledge. If you are going to have a tool analyze your unstructured text keep the underlying info in place.

Semantic Technology
In the beginning, files were linked
  ftp, Gopher, Archie
The the WWW linked groups of text and multimedia
  HTML and URLs
The future will allow individual objects of data to be interlinked
  People, places, locations, Events, etc., individually represented and linked into an integrated set of data.

Solutions:
Data can be browsed based on its attributes, also known as faceted search.

Queries across multiple systems, visualize, etc.
See:
http://www.environmentalhealthnews.com
http://otnsematicweb.oracle.com
http://www.lii.org

Search for previous designs that meet similar requirements. Search for reliability data, lessons, learned, and PRACA on a past part. This is one area that can be used to justify a ROI with a Sematic and Service Oriented Architecture.

Query example:
"Display the name, cost, reliability, year designed, and prevalence of fuel pumps weighing less than 10 lbs and capable of handling a viscosity of fuel > .9"

Semantic technology allows Ontologies to be merged by specifying equivalent high-level concepts. This is what the OWL language can be used for.

A vision of what Semantic Technology can do:
Traceability of engineering data
Rapid prototyping
Expertise locator
Improved reporting and exchange of data

Review: What is an Ontology (see slide 18)
The idea is to express a domain of knowledge in such a way that...

Semantic Technology in Government
The U.S. Government has made a strong commitment to adoption of semantic technology.

Getting started with Semantic Technology (See slide 21)

Discussion of standards. Complain that Microsoft is resisting ODF, but EU is forcing them to use it. (I mentioned that Notes 8 supports ODF.)

Next action: 1. Ask for Christopher for slides, 2. As for quick demo of his toolkit

Discussion/Comments (1):

Applying Semantic technology to KM

"Only once you have defined your ontology and use cases should you even begin to think about your IT infrastructure.

Need a common ontology - information about how to describe information"

This is where this stuff often falls down- defining a single ontology up front can be a recipe for failure and wasted time. See Clay Shirky's numerous articles on this topic. An optimal system will take an approach that allows for viewing the same data through the lens of multiple ontologies that change all of the time...

The company was called Inxight with an 'x', recently acquired by Business Objects.

Posted at 7/19/2007 8:05:41 AM by Matt McKnight



Discussion for this entry is now closed.