Tabular Topic Maps

1
Tabular Topic Maps

• Digital Camera Accessories TTM at Kodak
• Content Versioning
• Ontology Validation
• Tools

2
Need for the project

• Complexity was increasing
• The number of products offered for sale.
• The number of languages the product information
  was presented in.
• Granularity of information content variation
  across selling regions around the world.
• The relationships among various products.
• All of these requirements needed to be met with
  no increase in labor content.
• Excel should be used as familiar and ubiquitous
  across-the-company UI.
• Spreadsheets should be human readable. Learning
  curve should minimal.
• Spreadsheets should be sufficiently encapsulated
  and defined so that different departments could
  work on their own tasks independently.
• There should be an easy way to add new types of
  relationships without need to change existing
  data and processing flow.
• Information should be mergeable.
• There should be a defined process that turns
  merged knowledge into a website.

3
Progress

• Started with workbook design that was there.
• Optimized it and slowly moved to Tabular Topic
  Maps keeping UI design user friendly, Neutral
  syntax, hospitable to new types of relationships
• Sheets are independent and yet provide a clear
  interface for managing common sets of
  interconnected information objects.
• Requires one to think and organize content in
  terms of topics (products, regions, languages),
  vocabularies, associations, roles and contexts.
• The project started with a simple requirement to
  provide typical selling information relationships
  about Kodak's digital camera accessories and had
  triumphantly developed into a large content
  management challenge.
• Currently multilingual information and
  relationships between digital cameras,
  accessories and selling regions yield to over
  300,000 unique combinations on the Kodak.com
  website.

4
Full Cycle
5
Workflow
6
Why Controlled Vocabularies.

• Challenge
• Different teams work on various pieces of
  information independently and in parallel.
• How can results of their work merge?
• Agreements that allow unambiguous reference to
  topics of discourse.
• Should agree on the controlled vocabularies
• Should agree on the naming convention within
  controlled vocabularies
• Thereafter topics can be unambiguously identified
  by a unique combination of
• Controlled vocabulary
• Unique name in the scope of that vocabulary

7
Why Ontology

• An upper vocabulary should exist that covers
• Known control vocabularies
• Common classes of topics, topic occurrence
  (objects) , relationships between topics
  (associations) and role types played by topics in
  associations.
• Constraints
• There should exist common sense constraints
  imposed on topic characteristics in various
  contexts.
• There should exist a computer understandable
  schema of constraints correlations between topic
  characteristics.
• An agreement about possible relationships between
  products, accessories, OS, countries and other
  languages and about roles that various types of
  topics can play.
• TTM Ontology Rules for creating Subject
  Indicators constraints.
• Building Domain Ontology Modeling common
  grounds between parties.

8
Subject Identifiers and URN Notation in Excel
urnkodaklangde
urnKodakcategorymemory
urnkodakobjectproductinformation
urnkodakeknEKN006544
9
Subject Identifiers and URN Notation in Content
Files

• Metadata attribute on the content container uses
  the same URN notation to indicate the topic that
  it belongs to.
• 3877" urn"urnkodakcategorygear"
• Gear
• Keep your camera safe and secure when
  traveling with Kodak camera bags. Try our cables
  for connecting to your television or computer,
  too.
• Wherever you go, keep your digital camera
  safe and secure with a Kodak camera bag. We have
  three convenient sizes large, medium and small
  for every photographic need. Kodak also offers
  a wide range of cables for connecting your camera
  to computers or televisions.
• Spider creates a topic This matches with the urn
  used in the Workbook allowing the xsl that
  creates output to connect the content as
  needed.
• xlinkhref"urnkodakcategorygear"/ntity
• xlinkhref"urnkodakobjectcontent"/f
• klangen"/

10
Using Scalar Vector Graphics to Display
Relationships
Accessories availability by region.
11
Handling Content Version Changes in the Topic Map

• Requirement
• Minimize the number of generated files to be
  pushed to production.
• Use XSLT.
• Solution
• Compute topic map difference of the old and the
  new topic maps.
• Base difference computation on the underlying
  ontology.
• Use diff results to drive file generation for new
  and modified topics and delete instructions for
  deleted topics.
• Require every topic to have at least one subject
  identifier.
• Processing
• Use subject identifiers to find new and deleted
  subjects.
• Compare topic base names (scope).
• Compare topic occurrences (scope, type).
• Find new and deleted associations(scope, primary
  players)
• Compute new and deleted association role players.

12
Practical Ontologies for Tabular Topic Maps

• Inter-team slip-ups.
• A duplicated information
• A contradicting information
• Cross-team miscommunications.
• Examples
• New product was added by one team, but the
  marketing team did not add it to the
  recommended/compatible associations.
• New language was added but translators or content
  creators dead not realize it or lag behind.

13
Semantic Web Glasses for Topic Maps and RDF
http//www.cogx.com/swglasses
14
Processing
XML Topic Map
RDF Topic Map
Error Report

• XSL transforms Topic Map from XML to RDF
  representation.
• OWL Validator generates an error report based on
  OWL ontology referenced as topic and
  characteristics classes.

OWL Validator
XSLT Translator
15
Validating Topic Maps with DAML tools

16
Contextual Ontologies for Topic Maps

• Constraints are contextual.
• Set of classes can not be constrained without a
  context .
• The same topic can be an instance of
• A nice little boy in the context of his mother.
• A trouble maker in the context of teacher X.
• Set of classes can be constrained when the
  context is specified.
• A constraint can exist that in the scope of
  teacher X all students are one of
• Valedictorian,
• Perfect student (teachers pet in the scope of the
  rest of the class),
• Bright mind,
• Average student,
• Trouble maker,
• No explicit notion of context/scope in OWL

17
Modeling Ontology

• Find the contexts of the information to be
  modeled.
• Language, geographical, etc.
• Find and model common grounds between involved
  parties.
• Example A legal family may be constrained to
• Contain 1 man and 1 woman in the context of
  pre-200x USA,
• Contain any 2 spouses in the context of post-200x
  USA
• Contain 1 man and 1 or more women in some other
  parts of the world.
• Example There should be text-labels in every
  language context.
• Contexts may be expressed directly using scopes,
  or indirectly via topic characteristics patterns.
• Outcome
• Track such inconsistencies as missing or
  duplicate occurrences of required types, invalid
  associations between topics, etc.
• Some of the errors that we are able to catch are
  very tedious and normally shows up only upon
  reviewing a corpus of generated web pages.
  Automated error tracking provides a higher level
  of quality content for the real time page
  creation process.

18
Coherent Contextual Characteristics Constants.C4
Expressions
One set of topic characteristics constrains other
characteristics of a topic in a given context

• - add-hock Class description.
• - Class definition
• A topic is an instance of an add-hock class if it
  matches this pattern.
• - set of constraints
• An instance of this add-hock Class should also
  match all patterns from this set of constriants.
• and contain CCX topic
  queries.
• CCX Topic queries may reference add-hock classes
  defined elsewhere.

19
Processing

• XSL templates are turn C4expressions into a
  validating style sheet.

Error Report
Topic Map

• The validating stylesheet applied to a topic map
  generates a report displaying warnings and errors.

Topic Map Constraints
XSLT Validator
20
C4 Expression example (Kodak).

• /1.0/psi1.xtmat-class-instance"
• 1.xtmrole-instance"/
• psi1.xtmrole-class"/
• egory"/
• cardinality"1"
• cardinality"1"

21
Topic Map Ontology and Validation

• A sample of a few violations found in the
  operational topic map for the accessories site.
  Note we have both missing components and
  duplicate components.

22
C4-Expression Example 1.

• Example
• Pattern A class of all topics that have US
  Passports
• Restrictions Must have one and only one SSN

typeurnusssnnum cardinality1/

23
C4-Expression 2a

• A class of all topics that graduated XYZ
• Topic that plays "http//www.cogx.com/psi.xtm/rol
  e-student role in association of class
  http//www.cogx.com/psi.xtm/at-graduates where
  "http//www.cogx.com/psi.xtm/role-college role
  is played by urnuscollegexyz
• Must have at least one web page

at-graduates role
"http//www.cogx.com/psi.xtm/role-student"
com/psi.xtm/role-college"/ ref"urnuscollegexyz"/ onAssociation
eb-page minCardinality1
24
CCX Example 2b

• A class of all topics that graduated XYZ
• Must have a web page in every language taught at
  College XYZ

each"x" in"xyz-taught-langs"
web-page" cardinality"1" ref"x"/