A repository of online PDF copies of
many of the AIAI publications is available.
You might also like to try our
Projects Page (or
Pre-1999 Projects Page) for
pointers to more publicatons and information on our work.
1995 Technical Reports with Abstracts
-
AIAI-TR-206
"Artificial Intelligence for Project Planning"; Brian Drabble;
Proceedings of the Colloquium on Future Developments in Project
Management Systems; Institute of Electrical Engineers, Savoy Place,
London; October 1995;
Abstract
The aim of this paper is to outline some new ideas that could help
create the next generation of project management technology. This
paper has four main sections. The first of these defines the problem
to be addressed, the second outlines the strengths and weakness of
existing project management software tools, the third describes what
intelligent project planning could provide, and the fourth lists the
particular technology items that need to be developed to provide an
intelligent project planning environment.
-
AIAI-TR-193
"Development of a KBS for Personal Financial Planning Guided by
Pragmatic KADS"; Susanne Dirks, IBM Ireland; John Kingston, AIAI and
Mandy Haggith, University of Edinburgh; This article was published in
the journal "Expert Systems with Applications", 9,2, 1995, pp. 91-101;
Abstract
This paper outlines the structure and the results of a research
project titled Development of a KBS for Personal Financial Planning
Guided by Pragmatic KADS. The aim of the project was the
development of a prototype knowledge-based system for personal
financial planning. The prototype system developed assists a
potential private investor in selecting a professionally balanced
mixture of investment products that meets all the investor's
requirements with regard to the investment. As part of the
specification, it was decided that the prototype is to develop a
portfolio for a single client investing a one-off lump sum, assuming
that the client's pension, insurance and mortage have been arranged
and that no other prior investments need to be taken into account.
The development of the prototype was guided by the Pragmatic KADS
methodology, a streamlined version of KADS suitable to the development
of small and medium-sized systems. The paper therefore describes how
Pragmatic KADS helped to guide and structure the development of the
prototype. The development process showed that the models that are
developed as part of KADS knowledge analysis, which view knowledge
from different viewpoints, contribute both to breaking down the
complexity of the problem and to ensuring a comprehensive knowledge
analysis. It also shoed that the top-down approach of KADS has the
inherent danger of neglecting some of the low-level knowledge that may
have an impact on the high-level problem-solving strategies. This
paper presents a way in which this problem can be resolved by using a
modified version of the spiral approach to expert system development,
guided by KADS.
PDF file
-
AIAI-TR-190;
"KBST: A Support Tool for Business Modelling in BSDM"; Yun-Heh Chen-Burger,
David Robertson, John Fraser, Christine Lissoni; Applications and Innovations
in Expert Systems,
Proceedings of the 15th Annual Conference of the British Computer Society
Specialist Group on Expert Systems (SGES), Cambridge, UK, December 1995;
Abstract
This paper describes a knowledge-based support tool for business
modelling with IBM's Business System Development Method (BSDM). The
tool, KBST, is designed to support the activities and capture the
results of BSDM workshops, where senior business managers together
with a BSDM facilitator develop business models. In this paper, we
show how case-based reasoning techniques can be used to build domain
specific knowledge into such a tool and how this can provide guidance
in building appropriate business models. It is also shown that the
application development package HARDY, a hypertext diagramming tool,
provides a good platform for a BSDM support tool.
PDF file
-
AIAI-TR-189;
"A Profile of AIAI"; Ann Macintosh;
IEEE Expert: Intelligent Systems and their Applications, June 1995;
Abstract
This is the first paper in their special series that
features AI Labs from around the world. Each issue highlights one lab
that is devoted to applied research and technology transition. It
addresses the institute philosophy, technical strengths and focus, and
discusses specific technology transitions and their impact.
-
AIAI-TR-188;
"An Engineer's Approach to the Application of Knowledge Based Planning
and Scheduling Techniques to Logistics"; Austin Tate, Brian Drabble, Jeff
Dalton;
Abstract
O-Plan is a command, planning and control architecture with an open
modular structure intended to allow experimentation on, or replacement
of, various components. The research is seeking to determine which
functions are generally required in a number of application areas and
across a number of different command, planning, scheduling and control systems.
O-Plan aims to demonstrate how a planner, situated in a task
assignment and plan execution (command and control) environment, and
using extensive domain knowledge, can allow for flexible, distributed,
collaborative, and mixed-initiative planning. The research is seeking
to verify this total systems approach by studying a simplified
three-level model with separable task assignment, plan generation and
plan execution agents.
O-Plan has been applied to logistics tasks that require flexible
response in changing situations.
The O-Plan research has achieved a clearer understanding of the
components necessary in a flexible planning system, and has shown how
such components can be combined in an open systems integration
architecture. The work has determined improved ways in which the
knowledge available from modelling an application domain can be used
effectively to restrict search in a planner. An improved
characterisation of a plan as a set of constraints on activity opens
up many possibilities for richer distributed, cooperative and
mixed-initiative planning systems in the future. The project has
created a prototype implementation which has been demonstrated on a
class of realistic applications.
PDF file
-
AIAI-TR-187
see AIAI Project Report, AIAI-PR-63 "PROPOSE - PROject POrtfolio SElection"
-
AIAI-TR-186
"Neighbourhood Search and the Vehicle Routing Problem"; Tim Duncan;
Abstract
Distribution and logistics account for a sizable part of many
companies' costs and therefore methods for increasing efficiency are
much sought after. The Vehicle Routing Problem (VRP) abstracts the
main features of such problems and in this paper we describe an
approach, based on neighbourhood search techniques, which has been
applied to the VRP. We also investigate the effect of varying the
proportion of the solution population in generating new solutions, and
test the system on a number of benchmark problems.
-
AIAI-TR-185
"O-Plan's Plan World Viewers";
Austin Tate, Brian Drabble;
Proceedings on Fourteenth UK Special Interest Group on Planning and Scheduling, Wivenhow House Conference Centre, Essex ; 31/7/95.
Abstract
The user interface to the O-Plan planning system seeks to
differentiate the various roles played by users in systems which
support command, planning and control functions. Appropriate support
is offered to the Task Assigner, the planning specialist and the
operational execution staff.
The planning role is supported by a user interface that provides
different views of the plan structure. These can be technical or plan
structure oriented views, or they may be more visualisation or world
oriented views. We provide support to either view via an interface
that supports the "plugging-in" of appropriate PlanWorld viewers which
conform to a specfied interface.
PDF file
-
AIAI-TR-184
"O-Plan: A Situated Planning Agent";
Brian Drabble, Austin Tate;
Proceedings of the Third European Workshop on Planning (EWSP-95), Assisi, Italy,
September 1995; 31/7/95.
Abstract
This paper describes the need to view a planner as situated in an
environment dealing with the whole "planning" problem. While a
planning agent deals with plan generation aspects, other agents are
concerned with aspects such as task elicitation, plan analysis,
reactive execution, plan repair, etc. Each of these systems has its
own perspective on the planning problem and each of the systems must
be capable of communicating in a way which allows other systems to
assimilate new information into their perspective of the problem.
Within such a collection of agents a situated planner takes task
assignments from a superior agent and creates a plan or further
elaborates it before passing it to the execution support agents for
further processing or enactment. The paper describes the O-Plan
system which has been developed as an architecture within which
situated agents, such as planning agents, can be created. The paper
provides a summary of work to date on the planning and execution
agents. The paper then goes on to describe current research involving
O-Plan which aims to address the communication between a task
assignment agent and a planning agent. It concentrates on three key
issues in this area: communication of plans, assessment of the quality
of plans and the role of authority in the planning process.
PDF file
-
AIAI-TR-183
"Towards a Methodology for Building Ontologies";
Mike Uschold, Martin King;
Workshop on Basic Ontological Issues in Knowledge Sharing,
held in conjuction with IJCAI-95; 31/7/95.
Abstract
We outline some requirements for a comprehensive methodology for
building ontologies, and review some important work that has been done
in the area which could contribute to this goal. We describe our own
experiences in constructing a significant ontology, emphasising the
ontology capture phase.
We first consider the very general issue of categorisation in
modelling, and relate it to the process of ontology capture. We then
describe the procedure that we used to identify the terms and produce
definitions. We describe a successful way to handle ambiguous terms,
which can be an enormous obstacle to reaching a shared understanding.
Other important findings include: it may not be necessary to identify
competency question before building the ontology; the meta-ontology
can be chosen after detailed text definitions are produced; defining
terms which are 'cognitively basic' first can lead to less re-work.
PDF file
-
AIAI-TR-180
"Schedule IT: An Intelligent Vehicle Scheduling System";
Tim Duncan;
Proceedings of the first ILOG SOLVER and ILOG SCHEDULE International Users'
Meeting, Abbaye des Vaux de Cernay, France,; 30/6/95.
Abstract
This paper describes the experiences and insights gained during the
development of an intelligent vehicle scheduling system using ILOG
SOLVER. A prototype was described in an eariler paper (Duncan 94).
Here we look at some of the issues involved in the design of the
system, and improvements that were made for the current system.
The main task of the system is to allocate drivers and vehicles for
the delivery of orders to customers. Since each order generally
requires a separate journey, vehicle routing does not play a part in
the system.
Distribution constitutes a major proportion of the client company's
business costs, and the quality of the service that they provide
(mostly in terms of timeliness) is an important differentiator from
other companies in the same field. This lead to the twin aims of the
project being to improve the quality and consistency of schedules,
while reducing costs.
-
AIAI-TR-179
"Schedule IT: Improving Delivery Service through Constraint-based Vehicle
Scheduling";
Tim Duncan, Howard Beck;
Proceedings of International Conference on Transporation and Constraint
Programming, Montpelier, France, published by E; 30/6/95.
Abstract
Schedule-IT is a vehicle scheduling system developed by AIAI for an
oil and bitumen distribution company. The application involves the
assignment of drivers and vehicles for single-drop deliveries, under a
wide range of regulations, physical compatibility restrictions, and
operating preferences. The problem is therefore primarily one of
resource allocation rather than vehicle routing. The main function of
the scheduler to to produce predictive schedules for the next day, but
the system also provides support for the validation of user generated
or user modified schedules, and for reactive scheduling. The system
is split into two: the front-end which is responsible for maintaining
data, and the scheduler which calculates the detailed schedule. The
scheduler has been developed using ILOG SOLVER, a commercial library
for constraint-based programming.
-
AIAI-TR-178
"Intelligent Systems for Business Planning";
Brian Drabble, Howard Beck;
Proceedings of the AI in Business Planning and Monitoring Colloquium,
June 1995, IEE, Savoy Place, London; 31/7/95.
Abstract
Businesses face ever greater demands to change and improve. Many of
these demands are common across a wide-range of commercial and
government organisations. One critical element in the process of
improvement is the ability to generate and communicate business goals
and plans.
This report produced by AIAI for CCTA (the UK Government's Centre for
Information Technology) reviews the requirements of business planning
together with the intelligent tools and techniques which have been
developed to address this requirement. In particular, the report
shows how current intelligent business planning tools could be used to
address the challenges faced by H.M. Customs and Excise. The focus on
H.M. Customs and Excise case study is used to exemplify generic
business planning issues and to indicate potential solutions with wide
applicability.
-
AIAI-TR-177
"Yes, but why is that better";
Brian Drabble, Yolanda Gil, Austin Tate;
Proceedings of the International Conference on AI in the Petroleum Industry,
September 1995, Lillehammer, Norway; 31/7/95.
Abstract
Generating sets of qualitatively different plans from which a
selection can be made is crucial in the decision making support
systems within any organisation or business. Current business and
project planning tools are tasked such that all the alternative plans
or Courses of Action (COAs) generated are produced under some fixed
set of assumptions. The paper describes an approach which combines a
generative planner with a knowledge-based system that reasons about
plan evaluation criteria and allows qualiatively different solutions
to be generated. Examples of the use of these techniques in the area
of logistics planning are shown together with a summary of its
potential benefits in the oil industry.
-
AIAI-TR-176
"Experiments in the use of Neighbourhood Search techniques for Vehicle
Routing"; Tim Duncan;
Proceedings of Expert Systems'95 December 1995, Queens College,
Cambridge, UK; 30/6/95.
Abstract
Distribution is a substantial cost in many companies and in some
sectors it contributes a high percentage of the value added to goods.
Therefore the potential for savings is large, and there is much
interest in improving distribution efficiency. The Vehicle Routing
Problem (VRP) is an abstraction of a class of problems that has been
studied both as a theoretical problem and for its practical relevance
to applications. In this paper we investigate Neighbourhood Search
techniques for solving the VRP and describe some experiments with
variations on the approach applied to a set of benchmark problems.
-
AIAI-TR-175"A Distributed Scheduling Framework";
Carla Ludlow, Lyn Thomas, Austin Tate;
Proceedings of the Sixth International Conference on Tools with
Artificial Intelligence, pp 49-55, New Orleans, Lousisa; 31/5/95.
Abstract
A distributed problem solving approach to job shop scheduling is
described in this paper. The approach views the system as an
Organisation. Agents are assigned different roles and functions
depending on their position within the structure of the Organisation.
In this Organisation, agents of the same level state their interests
independently of each other and therefore Conflict is likely to occur.
A major thesis of the research reported here is that not only is it
important to deal with conflict but also that conflict as a
consequence of the scheduling process should be exploited as a way of
integrating different scheduling perspectives, as a way of allowing
agents to express their own interests independently of each other and,
thus, as a way of guaranteeing pluralism by providing agents with both
empirical knowledge (heuristics, dispatch rules and theoretical
knowledge (optimal algorithms).
-
AIAI-TR-174
"Open Planning, Scheduling and Constraint Management Architectures";
Howard Beck, Austin Tate;
Published in BT Technology Journal, January 1995, Vol 13 (1),
pp 95-101; 31/1/95.
Abstract
The development of open planning and scheduling systems seeks to (i)
support incremental extension and change, and (ii) facilitate
communication between processing agents (both computer and human).
This paper presents the open planning and scheduling approach adopted
in the O-Plan and TOSCA systems at the Artificial Intelligence
Applications Institute (AIAI) in Edinburgh. The purpose is to bring
together a description of the concepts developed at AIAI and to relate
them in a common framework. References to more detailed descriptions
are provided. The paper describes.
1. the key characteristics of the open planning and scheduling systems
developed at AIAI;
2. the basis of the separation of the constraint elements in planning
and scheduling tasks, distinguishing a high-level model of what
remains to be done, a user level view of the plan/schedule entities
and the low-level detailed constraints;
3. generic constraint managers for planning and scheduling including
time constraint, plan variables and resource constraints managers. An
example using the time point network within an activity or resource
reservation framework is provided.
PDF file
-
AIAI-TR-173
"Acquiring Criteria for Plan Quality Control";
Proceedings in AAAI Stanford Spring Symposium Workshop
on "Integrated Planning Applications", Stanford, University, Cali; 31/5/95.
Abstract
In producing plans, human planners take into account a variety of
criteria that guide their decisions. Besides constraints imposed by
the domain itself, these criteria often express preferences among
alternative plans that achieve the given goals. Human planners can
use these criteria for two important purposes:
when asked to generate one plan, human planners are able to discern
between an ordinary solution and a better quality one and propose the latter.
when asked to generate several alternative plans, human planners are
able to discern between similar alternative solutions and
qualitatively different ones relaxing different criteria to explore tradeoffs.
Current AI planners are good at generating a solution that satisfies
the goals that they are given. AI planners often have a facility to
use a plan evaluation function for the immediate needs of selecting a
good solution during the search for a single plan. However, they do
not usually integrate quality considerations across several plans.
The quality criteria that human expert planners consider.
.sp0.5
are highly dependent on the situation and the scenario at hand. Some
criteria may need to be considered if new considerations come up.
include complex factors and tradeoffs that are often not represented
by an automatic planner.
PDF file
-
AIAI-TR-172
"Characterising Plans as a Set of Constraints - The Model -
A Framework for Comparative Analysis";
Austin Tate;
Appears in the Special Issue on "Evaluation of Plans, Planners and
Planning Agents", ACM SIGART Bulletin, Vol 6, No 1, ; 31/5/95.
Abstract
The (Issues - Noes - Orderings/Variables/Auxiliary) Model is
a means to represent plans as a set of constraints. By having a clear
description of the different components within a plan, the model
allows for plans to be manipulated and used separately to the
environments in which they are generated.
PDF file
-
AIAI-TR-169
"Support for Managing the Dynamic Environment";
John Fraser;
Proceedings of the International Conference on Improved Manufacturing
Performance in a Distributed Enterprise: Advanced; 1/5/95.
Abstract
All companies, especially manufacturing ones, operate in a changing
environment. There are techniques, not yet widely used but put to
very good effect by some highly successful companies, which help to
manage the dynamic environment. Systems thinking and the scientific
approach are two of them. Fairly simple software tools exist to
support the techniques: when they can be integrated with information
management tools and more conventional modelling tools they will
combine to make powerful enterprise models. The Enterprise project is
exploring how to do this. Generic models, reuse and integration are
key issues for Enterprise.
-
AIAI-TR-165
"The Enterprise Tool Set - An Open Enterprise Architecture";
John Fraser, Austin Tate;
Proceedings of the Workshop on Intelligent Manufacturing Systems (IMS),
International Joint Conference on Artificial In; 31/8/95.
Abstract
The Enterprise Project is collaborative work between AIAI at the
University of Edinburgh, IBM UK, Lloyd's Register of Shipping, Logica
and Unilever. The project is establishing a generic framework within
which enterprise tools can be used to assist users in their tasks. It
is based on an Enterprise ontology which establishes shared
terminology for communication between users and tools. The paper
begins with an overview of the Enterprise Project and its current status.
The main aim of the paper is to describe the open systems integration
architecture which underlines the design of the Enterprise Tool Set.
This draws on earlier work in Open Agent-based Planning and Scheduling
Architectures (e.g., O-Plan [2], [8] and TOSCA [1]). A means to more
easily integrate new capabilities into the toolkit is provided through
the architecture which acts as a mediator between the various
components. This draws on earlier work on mediators which act to help
users of data base systems (e.g., PEXES/SDBA [4], [6]).
The core of the tool set will support user tasks via a workflow engine
which will assist the user in performing a task, allow access to
appropriate tools and methods, and make available suitable information
resources. An abstraction of this central workflow within the tool
set is provided in this paper. This acts to provide a framework for
describing the various components integrated within the tool set and
allowing them to be provided in a modular fashion [9].
PDF file
-
AIAI-TR-164
"MOBEDIC - A Decision Modelling Tool for Emergency Situations"; Jim Doheny and John Fraser; Expert Systems With Applications, Vol 10, 1996. Also published as AIAI-TR-164, AIAI, University of Edinburgh, April, 1995.
Abstract
This paper describes a software tool that we have developed at AIAI
for modelling the decisions that people make in emergency situations
in oshore environments. The tool was developed using C++ and runs on a
PC under MS Windows. It has a generic architecture and can be easily
extended to other environments with different characteristics, e.g.,
hospitals, commercial buildings, etc. We use frames to represent a
person's characteristics and their perception of the environment;
scripts are used to define typical behaviours for particular
situations. Our tool can be used to predict the likely behaviours of a
population in hazardous situations and help evaluate the effectiveness
of emergency procedures and training.
We have worked with our collaborators to integrate our decision model
with their model of people's movement to produce a system that can
realistically simulate emergency scenarios on offshore structures. We
believe that this is the first egress and evacuation modelling tool to
incorporate both decision making and movement modelling. Our work is
therefore an important step in the introduction of improved approaches
to the evaluation of offshore safety management.
Validating the decision model proved difficult because of lack of
suitable data. We acquired additional data by interviewing oshore
personnel and monitoring a mustering exercise. We then simulated an
oshore emergency scenario and the results were encouraging. In the
future we would like to enhance our model by incorporating
communication between personnel. This would allow us to model complex
scenarios, especially those that cannot be simulated realistically in
training exercises.
PDF file
-
AIAI-TR-160
"The Application of Knowledge Based Techniques to Support Authors and
Readers of Technical Documents";
Andrew Casson;
Proceedings of The Emerging Showcase '95, London School of Economics.
London, England, January 1995; 31/12/94.
Abstract
The production and use of technical and legal documents is an
important aspect of the day-to-day function of organisations, both in
commerce and government. However, the size, content, structure and
interactions of these documents can render them very complex, and as a
result the processes of creating and using them are difficult and
time-consuming
This paper proposes a knowledge-based approach to providing
intelligent support for the readers and authors of technical and legal
documents. This approach, which is based on rich hypertext
representations of documents are used correctly and efficiently,
and leaves the Interpretation of their contents to the reader.
Having outlined the technology, the paper goes on to discuss its
applications. In particular, a case study involving building
regulations is described. The application of intelligent
knowledge-based hypertext to this domain has been successfully
demonstrated using the PLINTH system designed and developed by AIAI
and the Scottish Office. Other applications discussed here include
design rationale and statute law.
Finally, the paper concludes with a summary and a discussion of why a
hypertext based approach is particularly appropriate at this time,
given the recent massive expansion of the global hypertext network,
the World Wide Web, and the exciting new opportunities for on-line
publishing that it heralds
-
AIAI-TR-159
"Intelligent Systems for Business Planning";
Drabble, B, Beck, H;
Proceedings of the "Emerging Technology Showcase 95", London School of
Economics , London, England, January 1995; 30/1/95.
Abstract
Businesses face ever greater demands to change and improve. Many of
these demands are common across a wide-range of commercial and
government organisations. One critical element in the process of
improvement is the ability to generate and communicate business goals
and plans.
This report produced by AIAI for CCTA (the UK Government's Centre for
Information Technology) reviews the requirements of business planning
together with the intelligent tools and techniques which have been
developed to address this requirement. In particular, the report
shows how current intelligent business planning tools could be used to
address the challenges faced by H.M. Customs and Excise. The focus on
H.M. Customs and Excise case study is used to exemplify generic
business planning issues and to indicate potential solutions with wide
applicability.
-
AIAI-TR-158
"Applying KADS to KADS: "
John K.C. Kingston, The International Journal of Knowledge
Engineering, 12, 1, pp. 15-26, February 1995 ; 1/2/95
Abstract
The KADS methodology [Schreiber et al, 1993] [Tansley & Hayball, 1993]
and its successor, CommonKADS [Wiellinga et al, 1992] have proved to
be very useful approaches for modelling the various transformations
involved between eliciting knowledge from an expert and encoding this
knowledge in a computer program. These transformations are
represented in a series of models. While it is widely agreed that
these methods are excellent approaches from a theoretical viewpoint,
the documentation provided concentrates on defining what models should
be produced, with only general guidance on how the models should be
produced. This has the advantage of making KADS and CommonKADS widely
applicable, but it also means that considerable training and
experience is required to become proficient in them.
This paper reviews three projects, which investigated the feasibility
of producing specific guidance for certain decisions which are
required when using KADS or CommonKADS to develop a knowledge based
system. Guidance was produced for the identification fo the generic
task addressed by a knowledge based system; for the selection of
appropriate AI techniques for implementing the analysed knowledge; and
for selecting a suitable tool for implementing the system. Each set
of guidance was encoded in its own knowledge based system, which was
itself developed with the assistance of KADS or CommonKADS. These
projects therefore both studied and applied KADS and CommonKADS in
order to produce knowledge based guidance for knowledge engineers.
The projects showed that it was feasible to produce heuristic guidance
which could be understood, applied and occasionally overridden by
knowledge engineers. The guidance provides reasonably experienced
knowledge engineers with a framework for making the key decisions
required by CommonKADS, in the same way that CommonKADS provides
knowledge engineers with a framework for representing knowledge. The
projects also produced some new insights about CommonKADS domain
modelling and about the process of task indentification.
PDF file
Last updated: 14th August 2015
by Austin Tate