Research Activities within the School of Computing

1
Research Activities within the School of Computing

• Welcome
• Aim
• To give an overview of our research activities
• To attract new research members and generate a
  synergy amongst colleagues
• Ethos of SoC Practical scholarship
• Last RAE assessment
• Information Systems (3A)
• Mobile Computing (3B)
• Applied Mathematics and Statistics (3B)
• Operational Research (2)

2
Research Seminars within the School of Computing

• A short introduction to our research activities
  but
• Research seminars (for more detailed description)
• All welcome to attend
• on Wednesdays 4pm (Amanda Quek
  A.Quek_at_staffs.ac.uk))
• Next research seminar
• Speaker Robert Buxton (CIES)
• Topic Fingerprinting oils, techniques allowing
  the development of a prototype pollution analysis
  and identification system
• Date 15 January 2003
• Venue Green Lecture Theatre
• Time 4-5pm

3
Schedule

• 1300-1430 Part 1
• Applied Maths Prof B. Burrows
• Stats and OR M. Fletcher
• Information Systems Prof H. Shah and her team
• IT Strategic Management Dr A. Eardley and his
  team
• Organisational Semiotics Dr R. Clarke and his
  team
• Artificial Intelligence Dr B. Sharp and her team
• 1440 1500 Tea/Coffee
• 1500 1700 Part 2
• Centre for Intelligent Environmental Systems
  Prof W.J. Walley and M. OConnor
• Mobile Network Systems Prof. R. Carrasco
• Graphics and Image Processing Dr C. Chibelushi
  and his team
• Web Engineering Dr L. Uden
• Plenary

4
Research in Artificial Intelligence

• Applications of AI techniques to process
  knowledge to solve specific problems
• Key themes
• Knowledge Based Systems (KBS), Agents Based
  Systems
• Summarisation, text mining, ontology and semantic
  web
• Members
• Staff Tim Barker and B. Sharp
• 2 PhD research students - FT (medical domain)
• Su Linying ( knowledge based image understanding)
• Vojtech Huser (clinical decision support systems)
  
• 2 Research Assistants (also P/T PhD students)
  TRACKER project (decision making process)
• Phil Windridge (ontology)
• Caroline Chibelushi (text mining)
• 2 PhD research students - PT
• Thomas Olsson ( concept formation), Desmond Case
  (agents)

5
Knowledge / Agents Based Systems

• Collaboration with STW and SWW (Rehabilitation of
  water mains)
• Began in 1997, over 50k
• Team Dr G. Bancroft, B. Edwards, B. Sharp (SoC)
  and A. Dean (Derby University)
• 14 papers, 1 PhD, 4 MSc projects D. Cases
  current PhD
• Problem
• Great pressure on the UK water industry to
  develop an effective rehabilitation strategy
• Aging water mains (corrosion and tuberculation
  problems)
• Wide distribution networks
• STW wealth of pipe data but inconsistencies and
  gaps
• SWW no data, apply experts knowledge and react
  to problems
• Need to respond swiftly to internal investment
  strategies, external policies, customer
  complaints, .

6
Research Issues

• STW SWW needed a system to integrate different
  types of knowledge
• Expert knowledge related to corrosion and pipe
  material
• Prediction of corrosion and life expectancy of
  water mains
• Strategic knowledge (for ex. investment policy,
  OFWAT rules)
• and to reflect the dynamic interaction(s) of
  these knowledge domains
• AI techniques being used data mining, neural
  networks, Bayesian belief networks, production
  rules, as well regression analysis (2 KBS)
• Research issues
• Can an agent based system provide a suitable
  platform?
• How can we represent, in an agent-based
  architecture, the dynamic interaction(s) of the
  various knowledge domains ?
• With how much autonomy should each agent be
  endowed?
• How can collaborative problem solving be
  effectively incorporated among agents?

7
Automatic abstracting

• Development of an Information Extraction System
  (INFORMEX)
• To generate abstracts from scientific papers
• Using Linguistic theories, Cognitive Science
  findings and AI techniques
• Extension of the method to French texts
• These techniques and in particular the idea
  behind Conceptual Constituents are being
  investigated further by Caroline Chibelushi in
  the TRACKER Project
• 1 PhD, 1 MRes, 3 MSc Projects, 11 papers

8
Other AI Research Projects

• Knowledge Based Systems in the Medical Domain
• Su Linying Knowledge Based Image Understanding
• Vojtech Huser Clinical Decision Support Systems
• TRACKER Project Working with IS and OS teams
• EPSRC grant in collaboration with Lancaster
  University
• Principal Investigator Prof A. Alderson
• Team B. Sharp, H. Shah, R. Clarke, 4 RAs
• Phil Windridge (RA) Representing Meanings in
  Ontologies
• Caroline Chibelushi (RA) Text Mining
• Collaborative Learning with Affective Pedagogical
  Agents to Aid Summarisation
• Tim Barker

9
Knowledge-Based Image Understanding by Spatial
Reasoning
Su Linying Supervisors
Bernadette Sharp Claude Chibelushi

• A knowledge-based approach to interpreting
  tibia/fibula X-rays in semantic description for
  fracture identification

10
Why knowledge-based approach?

• Image understanding needs knowledge
• Explicit knowledge representation helps not only
  machine but also human being
• Application expansion or modification without
  recoding

11
The problems

• Knowledge representation and integration (Spatial
  representation)
• Mechanism for utilizing knowledge (Spatial
  reasoning)

12
(No Transcript)
13
Knowledge Hierarchy
14
Spatial representation and reasoning

• Representing knowledge in Facts and Rules of
  expanded predicate logic (in fuzzy theory)
• Using Resolution Proofs in hybrid control as the
  reasoning strategy.

15
Clinical Decision Support Systems in Medicine

• Vojtech Huser, MD
• PhD, Sep 2002 - 2005

16
Clinical Decision Support Systems (CDSS)

• Focus of Medical Informatics
• Bioinformatics (DNA analysis, genome project)
• Bioimaging (image analysis)
• Clinical Informatics (CDSS)
• Public Health Informatics
• Researchers Medical Doctors, Computer Scientists

17
Clinical Decision Support Systems

• What are CDSS ?
• CDSS is a software that is designed to be a
  direct aid to clinical decision-making, in which
  the characteristics of an individual patient are
  matched to a computerized clinical knowledge base
  and patient-specific assessments or
  recommendations are then presented to the
  clinician or the patient for a decision (Sim,
  2001)

18
Clinical Decision Support Systems

• Why it is good to have CDSS?
• Number of medical facts is growing enormously
  (also general medical knowledge necessary for all
  doctors)
• Human brain capacity is limited
• Perfect cooperation among specialists within the
  health care system is needed
• Degrease the number of preventable errors (Kohn
  2002, Bates 2001)

19
Clinical Decision Support Systems project aims

• Investigate the existing CDDS and their pros and
  cons
• Global Systems (Iliad, DxPlain, QMR)
• Specialized System (Nephrology, Paediatrics,
  Acute Care, Cardiology, Gynaecology)
• Include new emerging trends in medicine (Evidence
  Based Medicine)
• Include new emerging standards for
• Clinical Guidelines (GLIF I and II)
• Electronic Patient Record (HL7)
• MeSH (Medical Subject Headings)
• UMLS (Unified Medical Language System)

20
Clinical Decision Support Systems project aims
(ctd)

• To make better use of existing medical knowledge
  systems (Medline, Isabel, include existing
  specialized systems as plug-ins)
• To suggest a new architecture for better global
  CDSS and to bring CDSS and its knowledge closer
  to practicing physicians
• More info at http//www.soc.staffs.ac.uk/vh4/mi/
  or v.huser_at_staffs.ac.uk

21
Representing meanings in ontologies

• Phil Windridge

22
Domain

• Systems/Software Development
• Various IT Professionals
• Analysts, designers, programmers, project
  managers, etc.
• Various user groups/types
• Strategic, tactical, operational

23
Ontologies
Wheeled Vehicle
Public Transport
Road Vehicle
Passenger Airplane
Commercial Aircraft
Bus
Commercial Passenger Airplane
Train
Adapted from http//opencyc.sourceforge.net/daml/c
yc-transportation.daml
24
Meanings
25
Representation
26
TEXT MININGCaroline Chibelushi13th Jan 2003
27
The project

• Another research activity of the TRACKER project
• To investigate the relationships between
  decision, decision making processes and rework in
  organisations using text mining techniques.
• Data a set of recorded transcripts of meetings
  and a set of written minutes of meetings.

28
The Text Mining Approach

• Aim is to discover information and relationships
  about decision making not contained explicitly
  within that collection
• Text mining make use of both Linguistic and AI
  approaches
• to analyse a collection of documents
  linguistically
• identifying sentences, lexical items, conceptual
  constituents
• some parsing (noun phrases, verb phrases,
  prepositional phrases)
• to identify an appropriate representation of the
  meaning of words, propositions and sentences in
  the text
• to apply data mining techniques to the documents
  to discover any patterns, associations and
  relationships hidden in the text

29
Text mining stages

• Discovery

Text
Text Segmentation
Text
Text
ion
Vector Representation
Patterns Discovery Relationships Associations
Patterns
Knowledge Validation
Discovery
Relationships
and associations
30
Expected Results

• To discover information and relationships between
  a decision, action, argument, issue and goal
• The information will be crucial in making
  decisions visible
• - Aim to reduce the number of decisions that
  necessitate subsequent rework

31
Collaborative Learning with Affective Pedagogical
Agents to Aid Summarisation

• Tim Barker

32

• Context CHALCS, Leeds
• A Level Physics course in a VLE
• Collaborative learning deeper engagement
• The Case of the Missing Peer ?
• Agent-based support for summarisation PROsila
• Animated Pedagogical Agents
• Opposing characters utilise affectations
• Agents evaluated at schools CHALCS
• Agents adapt affectively based on performance
• Future work Open Source software suite

33
www.timbarker.org