Ontologies for Knowledge Management - Eastern Mediterranean ...

collardsdebonairΔιαχείριση

6 Νοε 2013 (πριν από 3 χρόνια και 10 μήνες)

61 εμφανίσεις

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

1

Developing Ontologies for
Knowledge Management

Atilla ELÇİ

Dept. of Computer Engineering

Eastern Mediterranean University

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

2

Knowledge Management Topics


Motivation


Terms & Definitions


Roles of ontologies


PROTON ontology as bases for KM /
SemWeb Apps

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

3

Motivation


Knowledge Representation (KR):


= A world view: Building models of a
domain/problem which allow for automatic
reasoning and interpretation.


=> Formal semantics (Ontology!)


=> Machine
-
interpretable meaning


Semantic repository:


Storage, querying, and management of
structured data


DBMS vs Ontology
-
based


O
-
B provides depth of meaning not available through
DBMS

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

4

Terminology: KM views


What Is Knowledge Management

by the
The
Knowledge
Management Forum

(KMForum)
:


Read through these personal views on K & KM


Note the diversity of views & interests


Contrast & cross
-
check definitions of some viewers.



25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

5

Terminology


Dublin Core Metadata Initiative (DCMI, DC):
interoperable online metadata standards


Dataset: a set of structured data (list, table, DB,
etc.) useful for direct software processing


Ontology:


= Paradigm for KR in AI.


Conceptual schemata


Formal ontology as logical formalism as in OWL


Schemata or ‘inteligent’ views over information
resources:


For indexing, querying, and referencing non
-
ontological
datasets


For DB, Document Mngt Sys, Catalog, OLAP,

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

6

Terminology
(continued)


Ontology classification based on generality of conceptualization:


Upper
-
level ontology:


A general model suitable for large variety of tasks, domains, and
application areas. Can be used to line up independently developed
ontologies if linked to it.


Domain ontology:


For ‘specific’ domain of interest


App / Task ontology:


For a specific range of applications / tasks.


Knowledge base (KB):


A dataset with formal semantics and knowledge representation
allowing automatic inference.


Ontology: O=<C, R, I, A> where:


C: is the set of classes


R: is the set of relations among the classes


I: is the set of instances from the domain. Instances belong to classes


A: is the set of axioms (say, business rules).

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

7

Terminology
(continued)


Ref ontology definition as O
-
grammar, the issue of what is
instance what is schema definition may not always easily
resolved.


Data qualia: A data quale is an orthogonal quality of data that
may be used for independent classification:


Semantics: whether it is formally represented


Structure: if the data is formally structured


Schema: data that determines shape and/or meaning of ontology
data.


Sorts of data (“_” stands for ‘any value’ not determined):


Data: (_,_,_), ie. Any sort of collection of data


Dataset: (_,structured,_)


Knowledge Base: (semantic,structured,_)


Ontology: (semantic,structured,schema)


Non
-
semantic schemata: (nonsemantic,structured,schema)


Database: (nonsemantic,structured,schema)


Mixed datasets: (_,structured,schema&non
-
schema)


Content: (_,non
-
structured,_)


Metadata: data on data, annotation, ... How to represent in (?,?,?)?


Semi
-
structured data:

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

8

Terminology
(continued)


Sorts of data (continued):


Semi
-
structured data:


KR/NLP


Docs containing free text fragments in structured according to
some schema


DB


Data of non
-
relational data model.


Ref. Fig. 7.2
-

Structured vs semantic positioning of
various sorts of data.


25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

9

Roles of Ontologies


Ontology as Database Schema:


May not contain instance data.


Such as RDBMS schema.


Ontology as Topic Hierarchy:


Classification for various purposes:


DCMI and library classification


Yahoo
&
DMoz

taxonomies for Web data


See
Section 4 in this

for depth of Yahoo! Directory.


Compare Topic
-
Ontology versus Schema
-
Ontology (Sect.
7.5)


Ontology as Enterprise Resource Model:


Ref.: Ontolog
Database & Ontology

Mini
-
Series.

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

10

Mapping & Querying Disparate
Knowledge Bases


Self study: Davies
§
6.3

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

11

PROTON (PROTo ONtology) Ontology


A light
-
weight uppper
-
level ontology to serve as
model bases for information science community
for, for example:


Seed for ontology generation


Automatic entity recognition & information extraction


Metadata generation / semantic annotation.


Design Rationale:


For usage in KM & SemWeb appls


Light
-
weight: for being unrestrictive


Prefers not to deal with time & space


Low
-
cost of adoption & maintenance


Scalable reasoning

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

12

PROTON
(contiuned)

Consists of ~300 classes & 100 properties for:


Semantic annotation


Indexing, and retrieval.


Design principles:


Domain independence


Light
-
weight logical definitions


Alignment with popular metadata standards


Good collection of named entity types (people,
organizations, locations, numbers, dates, addresses.


Structure:


In OWL Lite


In four modules: System, Top, Upper, and Knowledge
Management (KM)


Organized á la DILIGENT Methodology,

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

13

PROTON
(contiuned)


Scope:


Developed in the
SEKT Project

through sampling of a corpus of
general news.


General entity types appearing commonly (Person, Location,
Organization, Money, Date, ...) are in PROTON Top.


KM aspects stems from:


KIMO of
KIM Project


OpenCyc


Wordnet


DOLCE


EuroWordnet


Voluntary compliance with:


Dublin Core


Automatic Content Extraction annotation types


Alexandria Digital Library

Feature Type Thesaurus


Future compliance with:
FOAF

and other popular standars &
ontologies.

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

14

PROTON
(contiuned)


Architecture:


Site at Semanticweb.org


Organized in three levels: Basic, Top, Upper


In four modules:


System (basic; protons:...): application ontology
meant for use by ontology
-
based software


Top (top; protont:...): abstractions


Upper (upper; protonu:...): specific cases


KM (upper; protonkm:...): specific cases

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

15

PROTON
(contiuned)


KM module: for application
-
specific extension of
PROTON:


Information Space:collection of themed info resources


Software Agent: specialized
Agent


User:
User

and
UserProfile


Profile


User Profile


Mention: name droppings, references to (private)
instances


Weighted Term: relates objects to numbers


Device: references to user devices.


Organizations


The
Knowledge Management Forum

(
KMForum
)


Virtual community of practice focused on furthering
fundamental theories, methods and practices. Features
archives and news.


What Is Knowledge Management


KM Forum


Boston
Knowledge Management Forum
:

A Community of
Practice
:

Learning and Working in the Knowledge Management
Community


KnowledgeBoard


Forum to establish a community and to support and identify
commonality in terminology, application and implementation.
Features news, workshops, a library,
...

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

16

Conferences


Knowledge Representation Ontology
Workshop (
KROW 2008
).



Eleventh International Conference on

Principles of Knowledge Representation
and Reasoning

(
KR 2008
),


Sydney, Australia, September 16
-

19, 2008

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

17

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

18

Commercial Conferences


Knowledge Base Publishing course series

of the Montague
Institute

includes articles:


Introduction to Knowledge Base Publishing


Taxonomies, search & Sharepoint


Metadata and search


Integrating taxonomies


Information modeling and metadata management



See also
Roundtables
, for example the following:


Benchmarking Sharepoint for KM

(December 12, 2007)


Six weeks to the Semantic Web

(November 7, 2007)


Integrating folksonomies with Google

(October 17, 2007)


Migrating metadata to the Semantic Web

(September 5, 2007)

25/04/'07 updated 15/04708

CmpE 588 Spring 2008 EMU

19

References


John Davies, Rudi Studer, Paul Warren (Editor
s
):
Semantic Web Technologies: Trends and
Research in Ontology
-
based Systems
, John Wiley
& Sons (July 11, 2006). ISBN: 0470025964
. Ch.
7.: pp. 115
-
138.


W3C
Semantic Web Tools

Wiki page:


Check ...