Introduction to Topic Maps and Subjectcentric Computing

1
Introduction to Topic Mapsand Subject-centric
Computing

• Bors István
• pepper.steve_at_gmail.com
• Budapest, 2009-07-21

2
Agenda

• Basic Concepts (TAO of Topic Maps)
• Advanced Concepts (scope and roles)
• Writing a Simple Topic Map in LTM
• Ontology-driven Editing with Ontopoly
• Breaks
• 10.30 11.00 (coffee)
• 12.30 13.30 (lunch)
• 15.00 15.30 (coffee)

3
The copernican revolution

• For 1,000s of years people thought that the sun
  revolved around the earth
• Actually some Greek, Indian and Muslim scholars
  knew better, but the view of Aristotle, Ptolemy
  and the Christian Church was dominant
• The publication of On the revolutions of the
  celestial spheres (1543) by Nicolaus Copernicus
  changed all that
• The heliocentric theory turned our understanding
  of the universe upside-down or inside out.

4
Computing has a similar problem

• Today we face a similar situation in computing
  and information management
• Our computing universe has applications (and
  documents) at the centre
• This is wrong, because it does not reflect how
  humans think
• Humans think in terms of subjects (or concepts)

5
The subject-centric revolution

• We must put subjects at the centre, because
  thats what really interests us
• For example, when looking for information
• This is the subject-centricapproach
• It represents a radically different way of
  organizing information and knowledge
• Subject-centric computing is what Topic Maps is
  really all about

6
What is Topic Maps?

• An ISO standard for computer-based
  informationand knowledge management
• Provides the ability to control infoglut and
  share knowledgeby connecting any kind of
  information from any kind of source based on its
  meaning
• A semantic technology
• Cf. Semantic Web (RDF, OWL)
• A form of knowledge representation
• Widely used for web-based delivery of information
• Plus Information Integration, eLearning,
  Business Process Modeling, Product Configuration,
  Business Rules Management, Asset Management,
  Knowledge Management,

7
Background to Topic Maps

• Emerged from the SGML community in 1990s
• Initial use case How to merge (digital)
  back-of-book indexes
• Some input from library science
• Precious little input from computer scientists
  before 2001
• Most of the SGML community came from the
  humanities
• ISO 13250 first published in 2000 (recently
  revised)
• A model for representing knowledge organization
  structures (indexes, glossaries, thesauri,
  encyclopedias)
• Plus interchange syntax, query language,
  constraint language, ...
• Widely adopted in Norway (esp. public sector)
• And gaining ground elsewhere

8
Basic ConceptsThe TAO of Topic Maps

• Topics
• Associations
• Occurrences

9
The TAO of Topic Maps
Callas, Maria 42 Cavalleria Rusticana
71, 203-204 Mascagni, Pietro Cavalleria
Rusticana . 71, 203-204 Pavarotti, Luciano
45 Puccini, Giacomo . 23, 26-31 Tosca
. 65, 201-202 Rustic Chivalry, see
Cavalleria Rusticana singers .
39-52 baritone . 46 bass
.. 46-47 soprano 41-42, 337
tenor . 44-45 see also Callas,
Pavarotti Tosca 65, 201-202

• The core concepts are derived from the
  back-of-book index
• Extended and generalized for use with digital
  information
• Consider a two-layer model consisting of
• a set of information resources (below)
• a knowledge map (above)
• This is like the division of a book into content
  and index

(INDEX)
knowledge layer
information layer
(CONTENT)
10
(1) The information layer

• The lower layer contains the content
• usually digital, but need not be
• can be in any format or notation or location
• can be text, graphics, video, audio whatever
• This is like the content of the book to which
  theback-of-book index belongs

information layer
(CONTENT)
11
(2) The knowledge layer

• The upper layer consists of (typed) topics and
  associations
• Topics represent the subjects that the
  information is about
• Like the list of topics that forms a back-of-book
  index
• Associations represent relationships between
  those subjects
• Like see also relationships in a back-of-book
  index

composed by
Domain Italian opera
composed by
Tosca
Puccini
MadameButterfly
born in
knowledge layer
Lucca
(INDEX)
12
Occurrences link the layers

• Occurrences represent relationships between
  information resources and the subjects that they
  are about
• The links (or locators) are like page numbers in
  a back-of-book index
• Occurrences canalso be typed (e.g.bio, map,
  synopsis)

13
Summary of core concepts
Lets look at some TAOsin the Omnigator

• The TAO of Topic Maps

Plus topic types, association types, occurrence
types each of which are represented by topics...
14
Omnigator interface
Demo
15
The power of the TAO model (1)

• Represent subjects explicitly
• Topics represent the things your users are
  interested in
• Capture relationships between subjects
• Associations provide user-friendly navigation
  paths to information (navigation as we may
  think)
• Associations promote serendipitous knowledge
  discovery through browsing
• Make information findable
• Topics provide a one-stop-shop for everything
  that is known about a subject (collocation of
  information and knowledge)
• Occurrences allow information about a common
  subject to be linked across multiple systems

16
The power of the TAO model (2)

• Represent taxonomies and thesauri
• Associations may represent hierarchical
  relationships
• Topic Maps permits multiple, interlinked
  hierarchies and faceted classification
• Transcend simple hierarchies
• Rich associative structures capture the
  complexity of knowledge and reflect the way
  people think
• Manage knowledge
• The topic map is the embodiment of corporate
  memory
• It provides a structured way to capture peoples
  knowledge of things, events, relationships, etc.

17
Querying topic maps

• Topic Maps is based on a formal data model
• This means that topic maps can be queried, like
  databases
• Topic Maps Query Language (TMQL)
• Allows more powerful use of taxonomies to
  retrieve information
• Permits queries that would make Google boggle
  (see below)
• Based on Ontopias query language tolog
• (Demo of querying in the Omnigator)
• Query example
• Give me all composers that composed operas that
  were based on plays that were written by
  Shakespeare

18
Advanced ConceptsScope and Roles
19
The problem of context

• A topic map captures knowledge, but...
• Some knowledge is only valid in a certain context
• Reality is ambiguous
• Knowledge has a subjective dimension
• People have different opinions
• Context is handled using scope
• Enables the expression of contextual validity
• Allows the expression of multiple world views

20
How scope works

• We make statements about topics
• names, occurrences, associations
• Every statement is valid within some context
• Statements are qualified by scope
• the name Allemagne for the topicGermany in the
  scope French
• a certain information occurrencein the scope
  technician
• a given association is true in thescope
  (according to) Authority X

21
Topics play roles in associations

• Associations have no direction
• They represent relationships andare inherently
  multidirectional
• Puccini was born in Lucca
• Lucca was the birthplace of Puccini
• Two ways to express the same relationship
• Impression of direction caused by use of natural
  language
• One of the topics viewed as the subject and the
  other as the object
• Instead of direction, associations use roles
• Puccini plays the role of person and Lucca plays
  the role of place
• person and place are association role types (or
  role types, for short)
• Labels are assigned based on role perspective

22
Anatomy of an association
person
born-in
place
T
T
T
T
R
A
R
T
Puccini
Lucca

• Role types characterize the nature of the
  subjects involvement in the relationship
• They are also topics

23
Associations need not be binary

• Unary associations are not common
• Useful for representing properties that have
  boolean values
• e.g., the property of being unfinished
• Binary associations are the most common
• Often correspond to verb ( subject, object )
  constructs
• Ternary associations are quite common
• Often correspond to verb( subject, direct-object,
  indirect-object ) constructs
• N-ary associations (where n gt 3)
• Less common but sometimes useful
• Many n-ary associations are better represented as
  (n-1) binary associations...

24
The Topic Maps standards

• ISO/IEC 13250 Topic Maps
• Part 1 Overview and Basic Concepts
• Part 2 Data Model
• Part 3 XML Syntax
• Part 4 Canonicalization
• Part 5 Reference Model
• Part 6 Compact Syntax
• Part 7 Graphical Notation
• ISO/IEC 18048
• Topic Maps Query Language
• ISO/IEC 19756
• Topic Maps Constraint Language
• ISO/IEC TR 29111
• Expressing Dublin Core Metadata Using Topic Maps

25
Creating a topic mapInterchange syntaxes

• HyTM, XTM, LTM and CTM
• Using LTM

26
Interchange syntaxes

• HyTM (HyTime Topic Maps)
• Original syntax, expressed in terms of SGML and
  HyTime
• No longer part of ISO 13250
• XTM (XML Topic Maps Syntax)
• Later, XML-based syntax, recently moved to
  version 2.0
• Easy to understand but very verbose
• LTM (Linear Topic Map Notation)
• Defined by Ontopia in 2001 and supported by other
  products
• A simple ASCII syntax for rapid prototyping
• CTM (Compact Topic Maps Syntax)
• ISO standard replacement for LTM
• Complete draft exists, but not yet finalized

27
XTM 1.0 Syntax example
lttopic id"la-boheme"gt ltinstanceOfgtlttopicRef
xlinkhref"opera"/gtlt/instanceOfgt ltbaseNamegt
ltbaseNameStringgtLa Bohèmelt/baseNameStringgt
ltvariantgt ltparametersgt
ltsubjectIndicatorRef xlinkhref"http//
www.topicmaps.org/xtm/1.0/core.xtmsort"/gt
lt/parametersgt ltvariantNamegtltresourceDatagtBoh
emelt/resourceDatagtlt/variantNamegt lt/variantgt
lt/baseNamegt ltoccurrencegt ltinstanceOfgtlttopicR
ef xlinkhref"homepage"/gtlt/instanceOfgt
ltresourceRef xlinkhref"http//www.opera.i
t/Opere/La-Boheme/La-Boheme.html"/gt
lt/occurrencegt lt/topicgt
28
LTM Syntax example
la-boheme opera "La Bohème" "Boheme"
la-boheme, homepage, "http//www.opera.it/O
pere/La-Boheme/La-Boheme.html"
29
LTM basics

• Topictopic-idpuccini composer
  "Puccini"lucca city "City"
• Associationassoc-type ( player role, player
  role )born-in ( puccini person, lucca
  place )
• Occurrencetopic-id, occurrence-type, "URL"
  topic-id, occurrence-type, string
  la-boheme, homepage, "http//www.opera.it/Op
  ere/La-Boheme/La-Boheme.html"la-boheme,
  premiere-date, 1896 (1 Feb)
• Scope(nameoccurrenceassociation) / topic-id

30
Demo Creating a topic mapwith LTM

• A simple knowledge management applicationto
  capture skills andexperience

31
What the topic map is about

• People are employed by organizations in certain
  professions.
• They have email addresses and other contact
  information.
• They are members of certain professional
  associations and they speak various languages to
  varying degrees.
• They attend various events (workshops,
  conferences) and write papers.
• Organizations have web sites and are located in
  certain cities

32
Some data

• Bognárné Lovász Katalin
• katalinbognarlovasz_at_gmail.com
• 36 305739349
• University of West Hungary
• Association of Hungarian School Librarians
• XI. Summer School for School Librarians
• School librarian and/or manager?
• Topic Maps Workshop
• Hungarian fluent
• English advanced
• German basic
• Fancy dress and tea in the school library(?)

• Horváthné Szandi Ágnes
• szandi_at_bolyai.nyme.hu
• University of West Hungary
• http//www.bdtf.hu/
• Szombathely
• Association of Hungarian School Librarians
• XI. Summer School for School Librarians
• http//www.ktep.hu/NYA2009
• Summer conference held every second year in
  different locations
• Association of Hungarian School Librarians
• Budapest
• http//www.ktep.hu/

33
Advanced ConceptsMerging and Identity
34
Merging topic maps

• Topic Maps can be merged automatically
• Arbitrary topic maps can be merged into a single
  topic map
• This cannot be done with databases or XML
  documents
• Merging enables many advanced applications
• Information integration across repositories
• Sharing and reusing taxonomies
• Automated content aggregation
• Distributed knowledge management

35
Principles of merging

• By definition Every topic represents exactly one
  subject
• Our goal Every subject represented by just one
  topic
• When two topic maps are merged, topics that
  represent thesame subject should be merged to a
  single topic
• When two topics are merged, the resulting topic
  has theunion of the characteristics of the two
  original topics

Merge the two topics together...
(Demo of merging in the Omnigator)
36
Subject identity

• Precondition for successful merging
• Knowing when two topics represent the same
  subject
• What makes merging possible?
• NOT the use of names, which are notoriously
  unreliable
• Names are not unambiguous (the homonym problem)
• Many topics have multiple names (the synonym
  problem)
• Achievement of the collocation objective
• Only possible through the use of unique global
  identifiers
• If subjects have unique identifiers, people are
  free to use whatever names they like, and topic
  maps can still be merged successfully

37
Subjects and Topics

• Topics are surrogates, or proxies (inside the
  computer) for the ineffable subjects that you
  want to talk about, such as Puccini, love, these
  slides, or the second law of thermodynamics

38
The identity of subjects

• Topics exist in order to allow us to talk about
  subjects
• The relationship between the two is sometimes
  called intentionality
• We need to know exactly which subject a topic
  represents
• That is, we need to establish its subject
  identity
• The collocation objective depends on knowing when
  applications are talking about the same thing

39
Subject identifiers

• The identity of most subjects can only be
  established indirectly
• An information resource can provide an indication
  of the subjects identity to a human
• Such a resource is called a subject descriptor
• A subject descriptor has an address,even though
  the subject it indicatesdoes not
• Computers can use the address of thesubject
  descriptor to establish identity
• Such addresses are calledsubject identifiers
• Subject descriptors and subject identifiers are
  the two sides ofthe human-computer dichotomy

40
Advice on subject identifiers

• Always use them for your typing topics
• Makes your topic map and your ontology more
  portable
• The more serious your application, the more
  extensively you should use them for instances
• Remember Merging with other topic maps will not
  be successful without identifiers
• LTM code for subject identifiers
• See ItalianOpera.ltm
• Example
• composer "Composer" _at_"http//psi.ontopedia.n
  et/Composer"

41
My conventions for PSIs

• URI prefix
• http//psi.ontopedia.net/
• Note Not all my identifiers have corresponding
  descriptors
• URI suffix
• Initial cap for topic types and role types (e.g.
  Composer)
• Lower case for association, occurrence and name
  types (e.g. born_in)
• Wikipedia conventions for instances
• Replace spaces with underscores
• Check Norwegian Opera for examples
• Do not use the Italian Opera Topic Map its
  conventions are outdated

42
Ontology-driven editing

• Creating topic mapsusing Ontopoly

43
What is an ontology?

• Shorter Oxford English Dictionary
• Ontology The science or study of being that
  department of metaphysics which relates to the
  being or essence of things, or to being in the
  abstract.
• Russell Norvig Artificial Intelligence
• A particular theory of the nature of being or
  existence
• Tom Gruber
• A specification of a conceptualization a
  description of the concepts and relationships
  that can exist for an agent or a community of
  agents
• Wikipedia
• A data model that represents a set of concepts
  within a domain and the relationships between
  those concepts
• John Sowa Knowledge Representation
• A classification of the types and subtypes of
  concepts and relations necessary to describe
  everything in the application domain

44
Topic Maps terminology

• Ontology
• the set of typing topics that is used within a
  given topic map, or that defines a class of topic
  maps
• i.e. the topic types, association types,
  occurrence types, etc.
• Constraints
• rules governing classes of objects (i.e. typing
  topics)
• Schema
• the combination of an ontology and constraints
• Schema language
• a language for writing schemas
• e.g. TMCL and OSL (Ontopia Schema Language)

45
Why you need an ontology

• An ontology in Topic Maps corresponds to
• the set of element types and attributes in XML
• the set of tables and columns in an RDBMS
• It determines the kinds of things that can exist
  in the topic map
• In other words, the ontology determines what you
  can say
• For example
• You cant express the fact that X and Y are
  organization unless you have a organization
  topic type
• You cant express the fact that person A is
  employed by organization B unless you have an
  employed by association type
• etc.

46
Expressing the ontology

• The ontology itself is part of the topic map
• Puccini is a topic of type composer
• Lucca is a topic of type city
• composer and city are also topics thatare
  present in the same map
• The association between Puccini and Lucca is of
  type born-in, where Puccini plays the role of
  person and Lucca plays the role of place
• born-in, person and place are alsotopics in the
  same map
• Lucca has an occurrence of type map and Puccini
  an occurrence of type bio
• map and bio are also topics
• Etc.

47
What is ontology-driven editing?

• A user-friendly way to create topic maps
• The equivalent of syntax-directed editing in XML
• The principle is simple
• The ontology describes what kind of things can
  exist in the topic map
• It also includes constraints on
• Which types of statement are used with which
  types of topics
• What cardinality they have
• Based on this, the interface is automatically
  configured for data entry
• The benefits
• Easier user interface no need to understand
  syntax
• More consistent topic maps
• Ontopoly is such an editor

48
How to use Ontopoly

• Read the Ontopoly User Guide!
• It will save you a lot of grief in the long run
• Start the program from OKS Samplers / Ontopoly
  Home
• Open an existing Ontopoly topic map
• Import an existing non-Ontopoly topic map
• Or create a new topic map
• Use the Description tab to describe the topic map
• (Also to validate it and a few other things)
• Use the Ontology tab to define the ontology
• topic types, type hierarchy, association types,
  role types, name types, occurrence types
• fields (names, identifiers, occurrences, and
  associations) that apply to each topic type,
  their order and cardinality
• Use the Instances tab to populate the data
• Uses an automatically configured forms-based
  interface

49
Some tips on ontology creation

• Sketch out the basic ontology on paper first
• Create the type hierarchy in Ontopoly
• Keep it simple
• Create association types and role types
• Specify what the role-playing topic types are
• Create occurrence types and name types
• Go to each topic type in turn, starting at the
  top of each type hierarchy, and assign additional
  fields
• Review the ontology
• Dont add data until you are fairly comfortable
  with the ontology
• Later changes to the ontology that invalidate the
  data may cause extra work

50
Some comments on Ontopoly

• Does not (yet) support scope or variant names
• Use typed names instead of scoped names
• Includes system information in the topic map
• The topic map can be exported without this
  information
• It can be hidden in the Omnigator
• Customize ? Nontopoly model
• Important points to remember
• Clicking on any link submits the HTML form, but
  does not save to disk
• You MUST click on the Save button regularly
• Changing the ontology when you have already
  entered data can lead to invalid data

51
Demo Creating a topic mapwith Ontopoly

• A simple knowledge management applicationto
  capture skills andexperience

52
Conclusion
53
Making information findable

• Intuitive navigational interfaces for humans
• The topic/association layer mirrors the way
  people think, learn and remember
• Powerful semantic queries for applications
• A formal underlying data structure
• Customized views based on individual needs
• Personalized information delivery using scope
• Information aggregation across systems and
  organizations
• Topic Maps can be merged automatically
• But there is more to Topic Maps than that...

54

• Today our desktops are application-centric and
  document-centric
• Icons represent applications and documents

55

• Why cant they be subject-centric, with icons
  that represent the subjects we are interested in?
• With links between related icons?
• And with context menus that allow us to find
  everything related to a particular subject?

gambia
K185
opera
topic maps
LING 2110
OOXML
tm2008
rana
INF 2820
janacek
bantu semantics
keynote
bayreuth
håkon
56
References (1/2)

• Articles
• The TAO of Topic Mapshttp//www.ontopia.net/topi
  cmaps/materials/tao.html
• ELIS article on Topic Mapshttp//www.ontopedia.n
  et/pepper/papers/ELIS-TopicMaps.pdf
• ISO standards
• http//www.isotopicmaps.org/
• Conferences
• International Topic Maps Users Conference
  (Oslo)http//www.topicmaps.com
• Topic Maps Research and Applications
  (Leipzig)http//www.tmra.de

57
References (2/2)

• Mailing lists
• http//www.infoloom.com/mailman/listinfo/topicmapm
  ail
• http//www.isotopicmaps.org/mailman/listinfo/sc34w
  g3
• Tools
• Overview of tools http//www.garshol.priv.no/tmto
  ols/
• Ontopia (Open Source Java engine)
  http//www.ontopia.net/
• Blogs, websites, etc.
• http//www.topicmap.com
• http//topicmaps.bouvet.no/blog/

58
(No Transcript)
59
Topic Maps and RDF

• Similarities
• Differences
• Interoperability

60
Semantic Web Layer Cake
61
Two households, both alike in dignity

• During the late 1990s the W3C and ISO developed
  two semantic technologies in parallel
• Two communities, largely unaware of each other
• Tackling the same fundamental problems
• Findability
• Semantic interoperability
• The results were RDF and Topic Maps

62
How the two families stack up
OWL
TMCL
RDF Schema
TMQL
SPRQL
QUERY
Topic Maps
ORG SYNTAX MODEL CONSTRAINTS
ORG SYNTAX MODEL REASONING
RDF
XML
CTM
XTM
LTM
RDF/A
RDF/XML
N3
ISOTopic Maps
W3CSemantic Web
63
Striking similarities

• Both extend XML into the realm of semantics
• Both allow assertions to be made about things in
  the real world
• Both define abstract, associative (graph-based)
  models
• Both have URI-based models of identity
• Both allow forms of inferencing or reasoning
• Both have XML-based interchange syntaxes
• Both have constraint languages and query
  languages
• But they are also different in some crucial
  respects...

64
Important differences

• Different roots
• Topic Maps has its roots in traditional finding
  aids (indexes, thesauri, etc.)
• RDF has its roots in document metadata and formal
  logic
• Different levels of semantics
• RDF is more low level Topic Maps has more
  higher-level semantics
• Different models
• Identity, scope, association roles, n-ary
  relationships, variant names,
• Different goals
• RDF An artificially intelligent web for software
  agents
• Topic Maps Findability and knowledge integration
  for humans

65
The Most Crucial Differences

• RDF/OWL is for machinesTopic Maps is for humans.

• RDF/OWL is optimized for inferencingTopic Maps
  is optimized for findability.

• RDF/OWL is based on formal logicTopic Maps is
  not based on formal logic.

• RDF/OWL is to mathematics asTopic Maps is to
  language.

66
Who can tell me what this is?

• Is it an H or an A?
• (Human or Agent)

• The point is that fuzziness is a fact.
• Humans can handle it machines cant.

67
Different capabilities

• RDF/OWL, to support logic-based inferencing,
  cannot allow fuzziness
• Topic Maps, because it is for humans, has to
  support fuzziness
• OWL ontologies tend to be very stringent and
  complex
• Topic Maps ontologies tend to be simpler and less
  formal
• OWL has properties for things that Topic Maps
  doesnt need
• Topic Maps has features that would be too complex
  for OWL
• So you need to decide what it is you really need

68
RDF or Topic Maps?

• RDF is more low-level oriented towards machines
• Topic Maps is more high-level oriented towards
  humans
• OWL is oriented towards artificial intelligence

• Do you simply want to encode document metadata?
• RDF is ideal and you wont need OWL
• Do you want to achieve subject-based
  classification of content?
• Topic Maps provides the best combination of
  flexibility and user-friendliness
• Do you want both metadata and subject-based
  classification?
• Go straight for Topic Maps because it also
  supports metadata
• Do you want to develop agent-based applications?
• Use RDF/OWL if you already have Topic Maps,
  youre half way there
• Whatever you choose, you can always move your
  data betweenTopic Maps and RDF, thanks to RDFTM

69
RDFTM

• RDF/Topic Maps Interoperability Task Force
• A task force within the Semantic Web Best
  Practices and Deployment Working Group
• Chartered to deliver two documents
• Survey of Existing Interoperability Proposals
• Guidlines for RDF/Topic Maps Interoperability
• Survey published in February 2006
• http//www.w3.org/TR/rdftm-survey/
• Draft guidelines published in June 2006
• http//www.w3.org/2001/sw/BestPractices/RDFTM/guid
  elines-20060630.html
• The task force is now disbanded and the work will
  be finalized by SC34

70
(No Transcript)
71
Applications of Topic Maps

• Taxonomy Management
• Metadata Management
• Semantic Portals
• Information Integration
• eLearning
• Business Process Modelling
• Product Configuration
• Business Rules Management
• IT Asset Management
• Asset Management (Manufacturing)

72
Taxonomy management

• For managing unstructured content
• Organization by subject because thats how
  users search
• A taxonomy is a simple form of topic map
• Topic Maps provides subject-based organization
  de-luxe
• Using Topic Maps offers many benefits
• Standards-based means vendor independence and
  data longevity
• Associative model allows for evolution beyond
  simple hierarchies
• The taxonomy can also be used as a thesaurus, a
  glossary or an index
• Identity model permits merging and reuse
• Dutch Tax and Customs Administration
  (Belastingdienst) uses Topic Maps as the basis of
  a taxonomy management system
• http//www.idealliance.org/papers/dx_xmle04/papers
  /04-01-03/04-01-03.html
• Capability can be added to any Content Management
  System

73
Metadata management

• A Metadata Server based on Topic Maps
• Management of metadata for government
  publications
• Used in the central public information portal
  (ODIN)
• Primary goal
• Ensure much greater consistency in the use of
  metadata across different government publications
  in order to improve findability for users
• ODIN now re-architected as regjeringen.no
• Solution based on Topic Maps

74
Semantic portals

• Topic Maps as the Information Architecture
• for web-based publishing (web sites, portals,
  intranets, etc.)
• Site structure is defined as a topic map
• Each page represents a topic (subject-centric)
• User-friendly navigation paths defined by
  associations
• Topics used to classify content
• Potential for subject-based portal connectivity
• Smooth evolution into Knowledge Management
  solutions

75
Enterprise information integration

• Topic Maps are designed for ease of merging
• Generate topic maps from structured data(or
  create topic mapviews of that data)
• Merge topic maps to providea unified view of the
  whole
• Easy to filter
• Create personalized viewsof this unified model
• Advantages
• Consolidated access toall related information
• No need to migrateexisting content
• Standards-based

76
Enterprise information integration

• Example Elmer project at Starbase (Borland)
• Integration server for software information
• Multiple disparate applications hold related data
• Unified topic map layer enables search across
  repositories
• Data integration without changing the underlying
  applications
• Portal interface
• Intuitivenavigation
• Full-text andstructured queries
• Smarttags integration
• Elmer terms (topic names)highlighted
• Provide links into theportal

77
E-learning BrainBank

• Topic maps are associative knowledge structures
• They reflect how people acquire and retain
  knowledge
• Students describe whatthey have learned
• Pilot users 11-13 year olds
• Key learning concepts are
• captured, named, described
• associated with other concepts
• Students are able to
• capture the essence of a subject
• describe what they have learned
• keep track of their knowledge
• Teachers are able to
• monitor students understanding

78
Business processes

• Multinational petrochemical company
• Uses TMs to manage business process models
• Flexible model allows arbitrary relationships to
  be captured easily
• Processes are modelled in terms of
• Steps involved, their preconditions, their
  successors, etc
• Processes related through
• Composition (one process ispart of another),
• Sequencing (one process isfollowed by another),
• Specialization (one process isa special case of
  a moregeneral process)

79
Product configuration

• Managing product configuration for mobile phones
• Products belong to families
• Features belong to products or product families
  and are grouped in feature sets
• There are dependencies between features and they
  apply in different regions, etc.
• Network of dependencies is already quite complex
• Now throw versioning into the mix!
• Managing all this data is not easy
• Dependencies modelled in a topic map
• Product configuration engineers use this to
  configureproducts using a very user-friendly
  interface
• System is driven by inference rules
• These work on the topic map
• Easily capture complex logic
• Also integrates with product documentation

80
Business rules

• US Department of Energy Rules for security
  classification
• Information about the production of nuclear
  weapons subject to thousands of rules
• Rules published in 100s of documents
• Most documents are derived from more general
  documents
• Guidance topics form a complex web of
  relationships
• Captured in a topic map (KB)
• Concepts connected to if-then-else rules
• KB used with inference engine
• automatically classifies information(documents,
  emails, ...), and
• "redacts" information (PDF, email, ...)
• Benefits
• Model expressive enough to capturecomplexity of
  the rules
• ISO standard stability longevity

81
IT assets

• University of Oslo Management of IT assets
• Servers, clusters, databases, etc. described in a
  TM (KB)
• Used to answer questions like
• If operating system Z is upgraded, what apps are
  affected?
• Service X is down, who do I call?
• If I take Y down, what else goes?
• Uses composite topic map
• Partly autogenerated
• Partly handcoded
• Two applications
• Whitney online
• Houston offline (foruse in emergencies)

82
Manufacturing assets

• US Department of Energy
• Topic map describes Y-12 manufacturing facility
• Provides overview of
• equipment,
• processes,
• materials required,
• parts already built,
• etc.

83
Tools (http//www.garshol.priv.no/tmtools/)

• ATop
• CmapTools
• ctm-mode
• dtddoc
• Escenic Topic Maps module
• Knowledge Concierge
• ltm-mode
• mappa
• OfficeNet Knowledge Portal
• Ontopia
• Perl TM
• QuaaxTM
• Ruby Topic Maps
• ThinkGraph
• tinyTiM
• TM/XMLtoXTM1 Converter
• TM
• TM4J
• TM4JScript

• TM4L
• TM4Web
• TMAPIX
• TMCore
• TMCore EPiServer Module
• TMCore Sharepoint Module
• TMCore Sitecore Module
• tmedit
• TMNav
• TMTab
• Topincs
• Wandora
• Wordpress Topic Maps
• xSiteable
• XTM1toXTM2 Converter
• xtm2xhtml
• xtm4xmldb
• ZTM

84
(No Transcript)
85
Topic types, type hierarchiesand other
hierarchies
86
Topic types

• A topic type defines a class of things
• Its a particular kind of category that has
  instances
• You can also think of it as a set of things that
  haveone or more properties in common
• Rule 1 If it doesnt have instances, it isnt a
  type!
• Music is a category, but not a type (there are
  no instances)
• nothing is a music
• Opera is a type, because there are things which
  are operas
• Tosca is an opera
• A diagnostic for deciding if foo is a type
• If you can think of things which are foos the
  answer is yes
• But be careful Is wine a type?
• If the answer is no, ask what kind of thing foo
  is
• Now, that really is a type!

87
ISA and type-instance

• The relationship between a type and its instance
  is actually a special kind of association
• We call it (guess what) a type-instance
  relationship
• Its also often called an ISA relationship
• It can be represented as an association in XTM or
  LTM
• But theres no real point
• Use the syntactic shortcut instead
• tosca opera

tosca
is a
opera
88
Rules of thumb for topic types

• Choose an appropriate level of generality
• Countries is better than Countries in
  South-East Asia
• The domain of the topic map tells you which
  countries it includes
• If it doesnt, an association would be a better
  solution
• located-in(Thailand, South-East_Asia)
• But dont make it so general as to be useless
• Places instead of countries would mix
  countries and cities
• Keep the name short
• That makes it easier to display
• Use the singular form
• Experience shows this to be most useful, so
  Country, not Countries
• Use initial capitals
• A matter of taste, but I think it looks most tidy

89
Type hierarchies

• Some topic types can be arranged in hierarchies
• Type hierarchies are a natural way to order parts
  of the world
• Humans are quite familiar with tree structures
• Type hierarchies provide
• more user-friendly navigation
• more powerful querying/inferencing
• more compact schemas and ontologies
• greater clarity about the relationships between
  types
• Use hierarchies, but beware of two pitfalls
• Not all hierarchies are type hierarchies...
• Its easy to confuse your ISAs and your AKOs

90
Type hierarchies AKO
a dog is A Kind Of canine, a canine is A Kind Of
mammal, etc.
91
Dragon 1 Mixing ISAs and AKOs
?

• Steve is a homo sapiens
• A homo sapiens is a mammal
• Therefore Steve is a mammal

• Steve is a homo sapiens
• Homo sapiens is a species
• Therefore Steve is a species

92
Types, subtypes and instances
93
How type hierarchies work

• The superclass-subclass relationship has defined
  semantics
• Therefore make sure you use it correctly
• Software (tolog, for example) will assume you
  mean what you say
• If you abuse the semantics you will get incorrect
  results!
• If A is a superclass of B, then
• Both A and B must be classes
• If C is an instance of B, it must also be an
  instance of A
• If C is a subclass of B, it must also be a
  subclass of A,(in which case an instance of C
  is also an instance of Band an instance of A)
• If in doubt define your own association type
• merging it with superclass/subclass later is
  trivial

94
Being both type and instance

• Most modelling paradigms distinguish between
  type and instance
• In most paradigms something cannot be both
• In Topic Maps something can be both type and
  instance
• (or class/category and individual)
• For example, homo sapiens can be both
• a type (supertypeprimate, instanceSteve), and
• an instance (typespecies)
• So be careful!

95
Representing a type hierarchy

• Use associations between typing topics
• subtypeOf(homo_sapiens subtype, primate
  supertype)
• subtypeOf(primate subtype, mammal supertype)
• XTM 1.0 defined identifiers for these three
  subjects
• subtypeOf (or superclass-subclass)http//www.top
  icmaps.org/xtm/1.0/core.xtmsuperclass-subclass
• supertype (or superclass)http//www.topicmaps.or
  g/xtm/1.0/core.xtmsuperclass
• subtype (or subclass)http//www.topicmaps.org/xt
  m/1.0/core.xtmsubclass
• Topic Maps software understands these and
  implements the semantics for you

96
Type hierarchies in LTM

• / Techquila hierarchy PSIs /
• hierarchical-relation-type "Hierarchical
  relation type"
• _at_"http//www.techquila.com/psi/hierarchy/hierar
  chical-relation-type"
• superordinate-role-type "Superordinate role
  type"
• _at_"http//www.techquila.com/psi/hierarchy/supero
  rdinate-role-type"
• subordinate-role-type "Subordinate role type"
• _at_"http//www.techquila.com/psi/hierarchy/subord
  inate-role-type"
• / XTM superclass-subclass PSIs /
• subtypeOf hierarchical-relation-type
• "Subtype of" "Supertype of" / supertype
• _at_"http//www.topicmaps.org/xtm/1.0/core.xtmsupe
  rclass-subclass"
• subtype subordinate-role-type "Subtype"
• _at_"http//www.topicmaps.org/xtm/1.0/core.xtmsubc
  lass"
• supertype superordinate-role-type
  "Supertype"
• _at_"http//www.topicmaps.org/xtm/1.0/core.xtmsupe
  rclass"
• / An example type hierarchy /
• subtypeOf( composer subtype , musician
  supertype )

/ Techquila hierarchy PSIs / hierarchical-relat
ion-type "Hierarchical relation type"
_at_"http//www.techquila.com/psi/hierarchy/hierarch
ical-relation-type" superordinate-role-type
"Superordinate role type" _at_"http//www.techquila
.com/psi/hierarchy/superordinate-role-type" sub
ordinate-role-type "Subordinate role type"
_at_"http//www.techquila.com/psi/hierarchy/subordin
ate-role-type" / XTM superclass-subclass PSIs
/ subtypeOf hierarchical-relation-type
"Subtype of "Supertype of" / supertype
_at_"http//www.topicmaps.org/xtm/1.0/core.xtmsuperc
lass-subclass" subtype subordinate-role-type
"Subtype" _at_"http//www.topicmaps.org/xtm/1.0/c
ore.xtmsubclass" supertype
superordinate-role-type "Supertype"
_at_"http//www.topicmaps.org/xtm/1.0/core.xtmsuperc
lass"
97
Dragon 2 Non-type hierarchies

• Not all hierarchies are type hierarchies
• For example
• geographical containment
• part of relationships
• subject classifications
• These relationshipsare not supertype-subtype
• located in
• part of
• subtopic of
• So again, be careful!

Norway is NOT a kind of Europe...
A piston is NOT a kind of submarine...
An opera is NOT a kind of music...
98
(No Transcript)
99
Topic Maps and Knowledge Organization

• Keywords controlled vocabularies
• Taxonomies, thesauri classifications
• Indexes glossaries
• Ontologies

100
Bibliographic languages

• Work language
• Author language
• Title language
• Edition language
• Subject language
• Classification language
• Index language
• Document language
• Production language
• Carrier language
• Location language
• Svenonius, Elaine (2000)The Intellectual
  Foundation of Information Organization.Cambridge,
  MA MIT Press (p.54)

• Work languages
• Work languages describe information entities,
  their intellectual (as opposed to physical)
  attributes, and relationships among them. (p.87)
• Document languages
• A document is a particular space-time embodiment
  of information a document language describes and
  provides access to this embodiment. (p.107)
• Subject languages
• A subject language is used to depict what a
  document is about. (p.127)

101
Two perspectives

• Works have tended to be conflated with documents
• So in practice there have been two kinds of
  language
• Document languages
• describe the work and its manifestations
• document-centric (or resource-centric), e.g.
• document metadata (Dublin Core)
• bibliographic records (MARC)
• Subject languages
• describe the subject space in which the work
  exists
• subject-centric, e.g.
• thesauri, taxonomies (ICD)
• classification schemes (LCSH, DDC)
• faceted classification (Colon)

102
Metadata

• Data about data
• Information about documents
• e.g. author, title, publisher, date, format,
  keywords
• Useful for managing the content
• Especially suitable for librarians
• Somewhat useful for searching
• Especially for experts
• Less useful for end-users
• the user starts out wanting to know more about a
  subject
• traditional metadata, however, focuses on the
  document
• if aboutness is provided at all, it gets squeezed
  into a single field

103
Keywords

• Primitive form of subject-based classification
• The keywords are used to describe the subject
• Cheap and simple Folksonomies and tagging.
• But also problematic because authors
• misspell keywrods,
• use different keywords/terms/tags for the same
  thing, and
• use keywords that make no sense
• Secondary problem
• No way for the user to find out what keywords
  have been used
• A keyword is a topic name

104
Controlled vocabularies

• Solution create a list of legal keywords!
• Requires somewhere to keep the list, and a
  process for new terms
• Benefits
• Solves problems of misspelling and duplicates
  (synonyms)
• Disadvantages
• Introduces some overhead (a flat list is
  difficult to manage)
• Users can still search using the wrong terms
• Users (and authors) still have difficulty finding
  terms
• A controlled vocabulary is a well-defined set of
  topics with one name per topic

105
Taxonomies

• Organize the keywords into a tree
• Most general at the top, more specific further
  down
• Common structure used by Yahoo!, etc.
• The folder metaphor
• file systems, email, favourites
• Requires relationships between terms
• Relationships state that one term is more
  specificthan another
• Advantage terms somewhat easier to find
• Disadvantage real world does not fit neatly into
  a hierarchy
• A taxonomy is a set of topics related through a
  specific type of hierarchical association

106
Thesauri

• Like a taxonomy, but with some extensions
• Also better defined there are ISO standards for
  thesauri
• Relationship types
• BT Broader term NT Narrower term
• USE Preferred term UF Non-preferred terms
• RT Related term
• SN Scope note
• A thesaurus is a set of topics related through
  particular, predefined association types
• BT/NT (hierarchical) and RT (untyped,
  associative)
• (Scope notes are a kind of occurrence)
• (USE and UF represent multiple names for the same
  concept/topic)

107
Faceted classification

• Invented by S. R. Ranganathan in the 1930s
• Defines a number of facets or dimensions
• Defines a set of terms within each facet
• Sometimes these terms are arranged in a taxonomy
• Documents are classified against each facet
  separately
• A faceted classification is a collection of topic
  hierarchies
• Each hierarchy contains topics whose names are
  used as terms within a particular facet
• XFML An XML interchange syntax for faceted
  classification inspired by Topic Maps

108
Expressivity progression
open model

• Topic maps and RDF/OWL
• use any types, properties, and relationships you
  like
• Faceted classification
• multiple vocabularies, taxonomies or thesauri
  (one per facet)
• Thesauri
• more formal taxonomy still no topic types two
  association types
• Taxonomy
• terms arranged in a hierarchy no topic types
  single association type
• Controlled vocabulary, folksonomies
• just a list of terms no topic types no
  associations

fixed model
no model
109
Document-centric approaches

• Traditional metadata is document-centric
• Provides substantial descriptive power for
  documents
• Allows connection into subject-based
  classification
• Crucial for the management of content
• However, users are most interested in the
  subjects
• Taxonomies, thesauri, and faceted classification
  are also document-centric
• These are methods for subject-based
  classification
• They provide hardly any descriptive power for
  subjects

110
Subject-centric approaches

• Topic maps are subject-centric
• They provide great descriptive power for subjects
• Good as finding aids, because subjects are what
  users care about
• Documents can be treated as subjects
• This enables topic maps to capture metadata as
  well
• It also enables topic maps to stitch metadata and
  subject-based classification together into one
  seamless whole
• Topic Maps is the knowledge model par excellence
• A subject-centric knowledge model that
  encompasses every other kind of knowledge
  organization model
• Topic Maps can therefore be used to relate and
  combine taxonomies, indexes, thesauri,
  classifications, etc. etc.