+ All Categories
Home > Documents > Hermann Helbig University Hagen Intelligent Information and Communication Systems Semantic Networks...

Hermann Helbig University Hagen Intelligent Information and Communication Systems Semantic Networks...

Date post: 01-Apr-2015
Category:
Upload: bailey-caton
View: 213 times
Download: 1 times
Share this document with a friend
28
Hermann Helbig University Hagen Intelligent Information and Communication Systems http://pi7.fernuni-hagen.de Semantic Networks as a Knowledge Representation Paradigm and Interlingua for Meaning Representation Part I - Foundations
Transcript
Page 1: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Hermann Helbig

University Hagen Intelligent Information and

Communication Systems

http://pi7.fernuni-hagen.de

Semantic Networks as a Knowledge Representation Paradigm and Interlingua

for Meaning Representation

Part I - Foundations

Page 2: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Semantic Networks[MultiNet]

* NL Access to Data Bases & to the Internet * Automatic Translation * Question-Answering Systems (QAS) * Semantic Web

Lexicography

Natural Language Understanding

KnowledgeProcessing

Helbig, University of Hagen Chair of Intelligent Information & Communication Systems

Cognitive Modeling

Logics

I. Disciplines:

III. Applications:

II. Technology:

* Workbench for the Knowledge Engineer * Natural Language Interpreter * Workbench for the Computer Lexicographer * Virtual electronic laboratory VILAB

Page 3: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

MultiNetKL-ONE

RMRS*)

*) Robust minimalrecursion semantics

The Position of MultiNet in the System of KRM

Page 4: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Basic Constructs

Page 5: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

The Upper Ontology(Sorts)

No collections, aggregations

as sorts

Page 6: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

The Layer Structure of MultiNet[Multidimensionality]

Page 7: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

„Max gave his brother several apples.“„This was a generous gift.“

„Four of them had been rotten.“

Intensional Level vs. Preextensional Level

Typical NL constructs aiming at the preextensional level:- apart from, the one ... the others, including, ....

Page 8: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Material Origin

o1 o2 p2 (o1 SUB o2) (p2 PARS o2) p1 (p1 PARS o1) (p1 SUB p2)D:

o1 o2 o3 (o1 SUB o2) (o2 SUB o3) (o1 SUB o2)K:

o1 o2 p2 (o1 PARS o2) (o2 ORIGM s) (o1 ORIGM s) D:

Sample Network

Page 9: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

B-Axioms

R-Axioms

Page 10: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Classification of axioms

Page 11: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Immanent vs. Situational Knowledge

Page 12: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Multilinguality

The Computer Lexicon[Semantic Part]

agent

orientation

beneficiary

result

mental content

Page 13: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Problematic Axioms and Rules of Standard Logic

Feasible Inference Rules of Logic

Guidelines for an Adequate Logic

- Principle of locality (admittance of contradictions)

- Opportunistic logic (fusion of different logical approaches)

- Associatively guided search (warranting a semantic coherence)

- Substitution of truth values by degrees of trustworthiness

A A B Extension Rule A A B Ex falso quodlibet ( A) A Law of Double Negation A A Law of the Excluded Third

Modus ponens Abduction Syllogisms

A B x R(x) H(x) A B A B x M(x) R(x)---------- ---------- ------------------------B A x M(x) H(x) (more trust- („ferio“) worthy)

Page 14: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

The Method of Question Centering

KB: „The travel agency T-Serv bought a PC from CompuTex in 1998.“Q: „Which firm did CompuTex sell a computer to?“

Page 15: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Hermann Helbig

University Hagen Intelligent Information and

Communication Systems

http://pi7.fernuni-hagen.de

Semantic Networks as a Knowledge Representation

Paradigm and Interlingua for Meaning Representation

Part II - Applications

Page 16: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

MWR –The Workbench for the Knowledge Engineer

Main Panel

Page 17: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Relation CAUS

The MWR Help System(Online Documentation)

Page 18: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

The Assimilation of Partial Networks

Ge: “Das Flugzeug mußte landen.“En: „The plane had to land.“

Ge: “Die Turbine wurde durch einen Vogel beschädigt.“En: „The turbine had been damaged by a bird.“

Ingo Glöckner

[REFER=det]

?

Page 19: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

The Result of the Assimilation Process

Ingo Glöckner

Inheritance

Page 20: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

<firm>

<publish>

<you>

<I>

<automatic>

<knowledge processing>

<book>

<person>

<processing>

WOCADI – The Natural Language Translator

Ge: „Nenne mir Bücher von Shapiro über automatische Wissensverarbeitung, die bei Addison-Wesley erschienen sind.“

En: „Show me books of Shapiro about automatic knowledge processing, which have been pulished by Addison-Wesley.“

Sven Hartrumpf

Page 21: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

The SQL Interface to Relational Data Bases

Johannes Leveling

Page 22: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

German Library

Result of the Search

Selection of Libraries.

Are there any books of Peter Jackson which have been published by Addison-Wesley.

Johannes Leveling

NLI-Z39.50 – A Natural Language Interface to the Internet

Page 23: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

GE: Gibt es Bücher von Peter Jackson über Expertensysteme, die bei Addison-Wesley erschienen sind? En: Are there books of Peter Jackson about expert systems which have been published by Addison-Wesley?   (QUERY = unspecified)(MEDIA_OBJECT = buch.1.1)(BIBCODE = (WORD b))(PUBLISHER = (NAME addison-wesley.0))((OR AUTHOR EDITOR) = (NAME jackson.0 peter.0))((OR TITLE (OR SUBJECT TITLE-SERIES)) = (WORD expertensystem.1.1))     find @and @attr 1=46 @attr 2=3 @attr 3=3 @attr 4=2 @attr 5=1 "expertensystem" @and @attr 1=1004 @attr 2=3 @attr 3=3 @attr 4=1 @attr 5=1 "jackson, peter" @attr 1=1018 @attr 2=3 @attr 3=3 @attr 4=2 @attr 5=100 "addison-wesley"

NL-Query

Interlingua Expression

YAZ Expression

Johannes Leveling

Page 24: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Textual Knowledge

Base

User searching for

duplicates

Recognizer

Data bases for texts

and for duplicates

Knowledge base

Theorem prover

Textual entailments

Semantic Recognition of Duplicates

Tim v. der Brück

Page 25: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Tim v. der Brück

DeLight – Linguistically Sound Readability Checker

Page 26: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

Max

cut down

ax

tree

Automatic Translation with MultiNet as a Semantic Interlingua

Max

fällen

Baum

Axt

Tiansi Dong

Page 27: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

SEMPRIA® GmbH

Answer field with ranked answers

Where did Oliver Bierhoff study?

Bierhoff finished his studies at the Open University in Hagen after ….?

Answer: at the Open University in Hagen? [Rank: +++]

SEMPRIA® GmbH

SEMPRIA® Search – A New Semantically Oriented Search Engine

Page 28: Hermann Helbig University Hagen Intelligent Information and Communication Systems  Semantic Networks as a Knowledge Representation.

SEMPRIA® Search – Comparison with Google

SEMPRIA® GmbH

Oliver Bierhoff gave homophobia a face ….?


Recommended