Language Technologies
Reality and Promise in AKT
Yorick Wilks
and Fabio Ciravegna
Department of Computer Science,
University of Sheffield
Overview
•
HLT
•
Using HLT for Knowledge Management
•
Challenges for HLT in AKT
–
Acquiring Knowledge
–
Extracting Knowledge
–
Publishing Knowledge
•
Demos
Human Language
Technology
•
Goal
–
Building systems able to process Natural
Language in its written or spoken form
•
Methodology
–
Use of Language Analysis
•
Technologies
(examples)
:
•
Information Extraction from Text
•
Human
-
computer Conversation
•
Machine Translation
•
Text Generation
HLT for KM in AKT
•
Use of HLT for
Acquiring
,
Retrieving
and
Publishing
Knowledge
•
Expected main benefits
–
Cost Reduction
–
Time needed for KM
–
Improving knowledge accessibility
•
Accessing/Diffusing/Understanding
•
Main challenges:
–
User factor
–
Integration
HLT in AKT
Knowledge acquisition retrieval publishing
Text mining
X
Information
Extraction
X
X
from Text
Classification
X
X
Summarization
X
Text Generation
X
Question
X
X
Answering
Traditional Knowledge Management
Drowning in information
Starving for Knowledge
Information Extraction
from Text
Question Answering
Text Summarization
Knowledge Management using HLT
HLT
Reports written
in natural language
•
Direct access to knowledge
when in textual format
?µ
Speed: Prompt Identification of
critical factors
?µ
Quantity: more information can
be accessed by people
?µ
Quality: only relevant
information is accessed by
people
?µ
Knowledge Sharing
University of Sheffield
Akt Challenges
•
Document classification
•
Text mining
Acquisition
Texts
Populating
with
instances
Extraction
•
Document classification
•
Information Extraction
Ontologies
•
Document Generation &
Summarisation
•
Agent Modelling
Publishing
HLT and KA in AKT
•
Use of text mining for:
–
Learning ontologies
•
taxonomies
•
Learning other relations
•
Main challenges
–
Integration of different techniques
–
Keeping track of changing knowledge
–
User factor:
•
interaction for setup and validation
Knowledge extraction
Information Extraction from Text
–
Populating ontologies with instances
•
Information Extraction from Text
–
Advantages:
•
Direct access to knowledge when in textual
format
•
Speed: Prompt Identification of critical factors
•
Quantity: more information can be accessed
by people
•
Quality: only relevant information is accessed
by people
•
Knowledge Sharing
Knowledge Extraction (2)
Question Answering
–
Retrieving knowledge from repositories
•
Question/Answering
–
Advantage:
•
Direct information access via Natural
Language
Q>
How
do
you
get
a
perfect
sun
tan?
NL
-
based
Question
NL
Answer
A>
Lie
in
the
sun
The user factor
•
Adaptivity for new application definition
–
Use of
Machine Learning
for new applications
•
Moving new application building towards non
experts
•
Time reduction
•
Criticality
–
The user factor in training the system:
•
What information/task can the user
provide/perform for adapting the system?
•
How can users know if the system does actually
what required?
Publishing Knowledge
•
Goal
–
getting knowledge to the people who
need it in a form that they can use.
•
Means:
–
Generation of texts from ontologies:
•
Knowledge diffusion
•
Knowledge documentation
–
Text summarisation
–
Generation of texts dependent on user
knowledge state
Knowledge diffusion
•
Advantages:
–
letting knowledge available:
•
In the form needed by each user
•
Expressed with the correct language type
•
Expressed with the correct level of details
•
Expressed without repetition of what is
known.
–
Skill reduction in querying ontologies
HLT infrastructure
•
KM requires a number of HLT techniques
to work together
•
Complex tasks require complex
interactions
•
Integration is then a main issue
–
How do you integrate the strength of each
technology to build an effective system
–
Working against current research paradigm
Conclusions
•
HLT provides many (potential) benefits
for KM
–
Effectiveness
–
Cost reduction
–
Time reduction
–
Subjectivity reduction
•
KM provides many challenges for HLT
–
User factors
–
Integration
Demo
•
Amilcare:
–
User
-
Driven Information Extraction from Text
–
Future Technology
–
Built in AKT
•
Trestle
–
Information Extraction
–
Current Technology
Thank You!
Enter the password to open this PDF file:
File name:
-
File size:
-
Title:
-
Author:
-
Subject:
-
Keywords:
-
Creation Date:
-
Modification Date:
-
Creator:
-
PDF Producer:
-
PDF Version:
-
Page Count:
-
Preparing document for printing…
0%
Σχόλια 0
Συνδεθείτε για να κοινοποιήσετε σχόλιο