Knowledge Technologies for Knowledge Management

1

• Knowledge Technologies for Knowledge Management

Dr. Rodrigo Martínez Béjar Universidad de Murcia,
Spain email rodrigo_at_dif.um.es
2
Contents of the presentation

• Ontological Engineering.
• Ontological Model.
• Knowledge Management.
• Applications of Ontological Engineering to
  Knowledge Management.
• Future Work.

3
What is an ontology?

• Meaning of existence
• In AI, if an entity can be represented then that
  entity exists.
• Explicit specification of a conceptualization.

• Conceptualization Structured interpretation of a
  particular domain.
• Set of concepts with relations between them.

4
Example
SCIENCES_FACULTY budget
PART-OF
PART-OF
PART-OF
PERSON age, income, role
BUILDING floors , height,
construction_date
KNOWLEDGE_AREA importance
IS-A
IS-A
STUDENT register_number, degree
STAFF dedication
IS-A
IS-A
PHD thesis_area
5
Concepts

• Ontological Knowledge Representation Basic Unit.
• All that can be represented.
• An ontology is comprised of a non-empty set of
  concepts.
• Attributes. Concept Properties. Specific Vs
  inherited. Uniqueness.
• Example

BUILDING
height,construction_date
6
Relationships

• They allow us to establish a conceptual hierarchy
  in the ontology.
• An ontology can be divided into subsystems
  according to the existing relationships.
• Examples of relationships.
• Mereology.
• Taxonomy.
• Topology.
• Temporality.
• Influence.

7
Axioms

• Rules for modelling conditions that must always
  hold.
• Types of axioms
• Structural. They derive from the existing
  relations between concepts and the existence of
  attributes. Example a IS-A b.
• Non structural. Relations between attributes.
  Example Fma.

8
Example
SCIENCES_FACULTY budget
PART-OF
PART-OF
PART-OF
PERSON age, income, role
BUILDING floors , height,
construction_date
KNOWLEDGE_AREA importance
IS-A
IS-A
STUDENT register_number, degree
STAFF dedication
IS-A
IS-A
PHD thesis_area
9
An Ontological model

• Elements.
• Concepts.
• Specific and Inherited Attributes.
• Relations.
• Taxonomic.
• Mereological.
• Temporal.
• Axioms.
• Structural.

10
Taxonomy

• Meaning Classification. A IS-A B --gt A is a
  class of B.
• Attribute inheritance. Taxonomic parents
  Specific Attributes.
• Properties Irreflexive, transitive and
  asymmetric.

PERSON age, income, role
IS-A
STAFF dedication
11
Mereology

• Meaning Science of theory of Parts. A part-of B
  --gt A is a part of B.
• Properties Irreflexive, asymmetric and non
  transitive.

SCIENCE_FACULTY budget
PART-OF
BUILDING height , construction_date
12
Types of mereologies

• Component/Object hand/arm.
• Member/Collection tree/forest.
• Portion/Mass piece of cake /cake.
• Characteristic/Activitypay/buy.
• Location/Area Murcia/Spain.
• Constituent/Object aluminium/plane.
• Phase/Process adolescence/development.

13
Temporality

• Meaning Temporal sequence of events or
  processes.
• Properties Irreflexive, transitive and
  asymmetric.

BULB_FAILURE
AFTER
BULB_REPLACEMENT
14
Axioms

• Based on the internal structure Type Concept
  has attribute.
• Derived from the properties of the relations
  between concepts. Type Concept A is not a class
  of Concept A.
• Temporal They establish temporal sequence
  between concepts.

15
Further work

• Relations.
• Types of partonomies.
• Topology.
• Non-structural axioms.
• Structured Attributes.

16
Ontological Integration

• Process of unification of the knowledge collected
  from a set of source ontologies.
• Main Apportions.
• Co-operative work is promoted.
• Individual knowledge is enriched by including
  others knowledge.

17
Knowledge Management (I)

• Control and Management of Organizational
  Knowledge to achieve the organization goal(s).
• Co-operative approach. Knowledge Sharing.
• High implication of the whole hierarchy of the
  organization (staff and directives).

18
Knowledge Management (II)
19
Applications of Ontological Engineering to
Knowledge Management

• 1) Corporate environments.
• Goal Implementation of corporate memories by
  using an Ontological Engineering based approach.
• 2) Medical environments.
• Goal Intelligent Alarm Systems for ICU by using
  an Ontological Engineering based approach.

20
Corporate Environments (I)

• Organization characteristics
• Strategy to achieve a goal.
• Resources .
• Intangible resources.
• Example Best practices.
• Inappropriate exploitation.
• Tangibles resources.
• Example Machinery.

21
Corporate Environments (II)

• The objective is to increase the performance of
  the organization by making a better usage of the
  available intangible resources.
• Corporate Memory. Explicit, intangible, and
  persistent organizational knowledge and
  information representation

22
Towards our goal

• Managing a corporate memory implies
• Creation Each member gives his/her knowledge.
• Distribution All the members must be able to
  access to the corporate memory.
• Use The corporate memory must be used.
• Maintenance Verifying that the knowledge of the
  corporate memory is valid, correct and
  actualized.

23
Application of the Ontological Model

• The previous tasks can be made by our ontological
  system
• Creation Each user sends his/her ontologies.
  Apportions Integration.
• Distribution The system users can access the
  global knowledge.
• Use The knowledge can be visualized.
• Maintenance Checking for the consistence of the
  memory and actualizing the ontologies.

24
(No Transcript)
25
Further work

• Adding new type of relationships
• Organizational environment.
• Higher realism.
• User facilities.
• Multimedia contents for facilitating knowledge
  accessibility and understanding.

26
Medical environments

• ICU.
• Alarm systems.
• Associated problems.
• Low specificity.
• Superfluous alarms.
• Objective Intelligent Alarm System.
• Superfluous alarms detection.
• Precise information about the event.

27
OBIAC
28
ICU Ontological Model
29
Application of the Ontological Model
Applied to

• Diagnosis Hypotheses formulation.
• Patients evolution.
• Checking diagnoses.
• Superfluous alarms detection.
• Selection of possible diagnoses.

30
Checking diagnoses
evolution
hypothesis
31
Superfluous alarms detection
hypothesis
evolution
32
Further work

• Interconnecting several ICUs.
• Multiple diseases management.
• Reuse of hypotheses.