[Ph.D. Position: Highly Interactive Object-Oriented Software Systems]

Rainer Blome rainer@physik3.gwdg.de
Wed, 21 Aug 1996 13:42:05 +0200


Hi folks,

the theme of this announcement shows some overlap with Tunes'
interface subproject.  Maybe someone is interested ...

Keep on rocking,       Rainer

------- Start of forwarded message -------
Return-Path: <stadt@cs.utwente.nl>
Date: Wed, 21 Aug 1996 13:15:35 +0200
From: stadt@cs.utwente.nl (Richard van de Stadt)
To: rainer@physik3.gwdg.de
Subject: Ph.D. Position: Highly Interactive Object-Oriented Software Systems
Content-Length: 8947

================================================================
Please forward this announcement or the URL containing the same
text - http://wwwtrese.cs.utwente.nl/Docs/phd/index.html,
to any qualified candidates you know who may be interested in 
applying to work in this project.
 
Thank you for your help,

Richard van de Stadt
TRESE project
Dept. of Computer Science
University of Twente
The Netherlands
================================================================

   We are currently searching for a Ph.D. candidate (F/M)
                      in the area of


    HIGHLY INTERACTIVE OBJECT-ORIENTED SOFTWARE SYSTEMS

INTRODUCTION

During the last decade, interfaces to software systems  have
evolved from low-level command interfaces to highly interac-
tive multimedia environments. Research  activities  on  mul-
timedia  environments  cover  many disciplines such as human
psychology, software engineering, media processing,  operat-
ing  systems,  computer  hardware,  databases, communication
protocols, media understanding, and entertainment.  In  this
project,  we  focus  on  the software engineering aspects of
designing interactive software systems.  In  particular,  we
will   focus  on  object-oriented  frameworks,  methods  and
models.

One of the major problems of interactive systems of today is
that  most systems provide only a static interaction pattern
with no opportunity for user interaction other  than  moving
from  window  to  window.  These interfaces can be termed as
"brochureware" because basically what  they  provide  is  an
"electronic  brochure" of the system. For example, in world-
wide-web  applications,  it  is   widespread   practice   to
translate  standard brochures into Hypertext Markup Language
(HTML) and putting them on the web.

Systems with static interaction patterns have  a  number  of
drawbacks.  Usability  of  these  systems  is limited to the
ability of users to browse and activate system  actions.  We
consider  "the  power  to listen" as a crucial feature of an
interactive system to improve usability and make it  attrac-
tive to its users. To be able to listen, interactive systems
must be equipped with the following  capabilities.  Firstly,
interactive  systems  must be able to "hear" what users have
to say.  Secondly,  systems  must  "understand"  the  users'
needs.  Thirdly,  systems must "respond" to users to satisfy
their needs. Finally, systems must learn about the  interac-
tions  that they have had with the users. We will term these
applications as "live applications".

Designing live applications  requires  the  following  tech-
niques:

1)   Since live applications are likely to  be  continuously
     changing,  they  require adaptable and dynamic language
     models for realization. We think  that  reflective  and
     composable object-oriented models are suitable for this
     purpose. In addition, interactive applications  require
     languages  that  are  capable  of expressing active and
     real-time objects with constraint enforcement features.

2)   Live applications are  not  restricted  to  information
     retrieval  only. They can be a part of any complex sys-
     tem, from electronic  banking  to  computer  integrated
     manufacturing.   Therefore,  design  methods  for  live
     applications must be integrated  with  general  purpose
     design  methods.  We think that current object-oriented
     methods have to be extended for this purpose.

3)   To listen, understand and  respond,  live  applications
     must be equipped with natural language processing tech-
     niques.  This  requires  extensions  to   methods   and
     language models.

4)   The developed language and method has to be  integrated
     with (de-facto) standard multimedia environments.


DESCRIPTION OF THE WORK

This research project is a joint activity between the  TRESE
and Parlevink projects, and aims at defining object-oriented
models and methods for building highly-interactive  software
systems. For this purpose, the following sub-goals are iden-
tified:

1)   Defining a composable, concurrent and real-time object-
     oriented  language  with constraint enforcement facili-
     ties for the construction of  live  applications.  This
     work  will  extend  the  composition-filters  approach,
     wherever necessary.

2)   Defining object-oriented methods to develop live appli-
     cations.  For  this  purpose, we plan the carry out the
     following activities:

     (a)  Defining object-oriented  software  artifacts  (or
          agents). Each artifact represents a design concept
          of a live application;

     (b)  Enhancing software artifacts  through  fuzzy-logic
          based design rules and design processes;

     (c)  Making artifacts programmable to represent partic-
          ular design information;

     (d)  Defining algorithms to match users' wishes and the
          design goals of the artifacts;

     (e)  Interfacing  the  software  artifacts   with   the
          scripting language of a hypermedia system;

     (f)  Integrating the language and method within a  mul-
          timedia environment.



CANDIDATE DESCRIPTION

First, the Ph.D. candidate must have a very good understand-
ing of the basic object-oriented concepts, methods, patterns
and frameworks. In addition, the candidate  is  expected  to
have  some  practical experience in object-oriented software
development. Second, knowledge on object-oriented reflection
and  concurrent  processing techniques are preferred. Third,
the candidate must have some knowledge about modern software
engineering environments, artifacts and process programming.
Fourth, knowledge about language understanding techniques is
highly  appreciated.  Finally,  the candidate must display a
high degree of eagerness and creativity necessary to produce
an excellent Ph.D. work.


EMPLOYMENT

The candidate will be employed by the University  of  Twente
as  a  Ph.D.  associate  based  on standard regulations. The
expected time span for the Ph.D. work is four years.


APPLICATION

If you are interested in this  position,  please  send  your
letter  including your C.V. not later than October 15, 1996,
with a list of courses/grades you have taken, to

Ms. Charlotte Bijron, E-mail: bijron@cs.utwente.nl, Depart-
ment  of  Computer  Science,  University of Twente, P.O. Box
217, 7500 AE, Enschede, The Netherlands. E-mail applications
are welcome.

To obtain more information, you can contact Mehmet Aksit:
(aksit@cs.utwente.nl)
or, Anton Nijholt:
(anijholt@cs.utwente.nl).


OUR RELATED BACKGROUND (http://wwwseti.cs.utwente.nl)

The TRESE Project

Since 1987, the TRESE  project  has  been  involved  in  the
development  of a large number of pilot applications to form
its  own  view  on  object-oriented  methods.  Although  our
experiences  confirmed  the  claims  about  the  benefits of
object-oriented methods, we identified a number of important
obstacles  that  are  not  addressed  by the current methods
[Aksit&Bergmans,  OOPSLA'92].   To  address  the  identified
problems,  we have introduced a new concept, called the com-
position   filters   based   object-oriented   model.    The
composition-  filters model aims at improving the expression
power of the object-  oriented  model  through  modular  and
orthogonal extensions.  Our major research contributions can
be summarized as improved object- oriented modeling [Aksit &
Tripathi,  OOPSLA'88],  reusable  transactions [Aksit et al.
IEEE Software March 91], reusable concurrency and synchroni-
zation  [Bergmans  &  Aksit,  JPDC, September 96], language-
database integration [Aksit et al.,  ECOOP'92],  abstracting
coordinated  behavior  [Aksit et al., LNCS791, 93], reusable
real-time systems [Aksit et al., ECOOP'94, Bergmans & Aksit,
JPDC   September   96],  fuzzy-logic  based  object-oriented
software development techniques [Aksit & Marcelloni, Metrics
Workshop, ECOOP'95].

The PARLEVINK Project

Within the Parlevink project a number of research activities
have  been  carried out in designing natural language inter-
faces for software systems and in designing  algorithms  for
several  kinds  of applications using neural networks, fuzzy
logic and genetic algorithms.

The Schisma subproject aims at designing a natural  language
accessible theatre information and booking system [G.F. v.d.
Hoeven et al., in  Proceedings  of  the  First  Workshop  on
Applications  of Natural Language to Data Bases, 95] [Ander-
nach et al., in  Language Engineering'95, AI'95, 15th Inter-
national  Conference,  95].   The  project  is a vehicle for
research on (multi-modal) dialogue modelling and  management
in human- computer interaction. Use is made of computational
models from theoretical computer science, including  machine
learning,   for  prediction  and  recognition  of  sequences
(words, categories, utterances, speech acts, dialogue  turns
etc.).


------- End of forwarded message -------