KASBAH_Proposal1.html
WEDNESDAY, 27 JUNE 2001
KNOWLEDGE ANALYSIS SYSTEM
BENEFITING ANIMAL HUSBANDRY
(KASBAH)
CONTENTS

OVERVIEW OF DOCUMENT

THE PROPOSAL

MOTIVATION

COMMITMENT

CONCLUSION

CONTACT INFORMATION

BIBLIOGRAPHY

OVERVIEW OF THE DOCUMENT

This document contains my project proposal for part-time research for an MPhil (with the potential to go on to do a PhD). It presents an initial view of the project, covering both the behavioural and computing aspects. I have also included sections on motivation and commitment to that project both on my part and on that of my employers, the Central Laboratory of the Research Councils (CLRC) at Rutherford Appleton Laboratory, Oxfordshire.

In brief, the proposal is to build a system that takes as input the conditions in which an otter in the wild lives in an optimal situation, to combine this with information about what constitutes a "happy animal" (as derived by various animal behaviour studies), and then use this information to provide a data/knowledge corpus that can also provide practical suggestions as to how to manipulate the variables that a zoo or wildlife park can control. The overall aims being to systematise our current information and to address issues of animal welfare: the keeping of captive animals in optimal conditions.

THE PROPOSAL

SYNOPSIS

This is an multi-disciplinary project combining elements of animal behavioural studies and animal husbandry (including aspects of animal welfare) with leading-edge computing techniques. The major proposed deliverable is a usable, information system with a Web front-end to assist those working with captive animals to improve both the levels of husbandry and of welfare for the animals in their charge. Spin-offs for the animal behaviour component of the project will the creation of an integrated information source (together with the necessary mappings and metrics). Although the underlying computer technology will rely upon well-proven data-base techniques, such features as the "inference engine" will also provide a clear research component for the computing aspect of the project. The resulting framework will allow the convenient addition of new information from laboratory experiments, studies in the wild and experience from animal husbandry. The proposed animal for the work is the otter (with a focus on the Asian Short-Clawed Otter, Amblonyx cinereus), although the proposed system architecture should facilitate the use of this approach with information relating to other species.

STARTING POINT AND APPROACH

The basic raw material for the work is a number of existing datasets relating to animal behaviour, physiology and pathology in a number of habitats (various wild environments, zoos, laboratories and other captive settings). The main characteristic of these datasets is that they are heterogeneous (often using very different, unrelated measures) and, to date, there have been few systematic attempts to integrate the information and extract knowledge that could be vital to good management both in the wild and in captivity. The proposed study species is the Asian Short-Clawed Otter (Amblonyx cinereus), but the techniques should be easily extensible to other species. There appear to be gaps in the current information, so part of the work would involve working, in the main, with zoos and the like, to fill these lacunae and to derive a suitable model. Then the use of meta-analysis and inference engines, combining behavioural and physiological indicators of a "happy animal" would produce a tool that would be applicable in the context of conservation and husbandry. The underlying data-resources and the support tools would typically be accessed via a Web front-end easily used by non-IT specialists. From my initial examination of the literature in the area, this would be an original and useful piece of research work. Drawing together the data in this way would also be an initial example of how such developing UK support tools emerging with the GRIDs ("e-Science") initiative could be used in a bio-medical domain.

INITIAL ARCHITECTURE FOR KASBAH IT SYSTEM

The KASBAH system will, for the pilot domain of the otter, bring together a number of datasets from the literature. These will be from disparate sources: studies in the wild, laboratory experiments, associated animal behavioural bodies and information from animal husbandry and welfare. The data will also be varied, ranging from almost anecdotal, single-case observations through to fully controlled clinical trials. There will thus be a data-collection, coding and integration phase to the project. Given the likely disparity of the data-types some form of meta-analysis (with associated schema) is likely to be needed. To fully integrate the information into a knowledge source an inference engine, SOUKH (System Organising Underlying Knowledge Handling) will be developed to make use of this integrated data-source, incorporating such techniques as "data mining" for particular queries. Discussions with colleagues researching in advanced database and artificial intelligence/KBS have suggested that the system may well lend itself to data mining using rule induction: observation -> conclusion. There would also seem to be scope for case-based reasoning (single-case; anecdotal evidence) and rule-based reasoning as possible ways of integrating different problem-solving and reasoning paradigms.

The usability of the system will be enhanced by a Web front-end for delivery to end-users (many of whom may not be highly "computer literate").

The diagram below (Figure 1) shows the initial concepts and components of the proposed IT support system:

Main Diagram of KASBAH System
FIGURE 1: Initial Diagram of KASBAH, Illustrating its relationship to the real world.

THE CHALLENGES FROM ANIMAL BEHAVIOUR

Three broad areas of work can be identified from the literature:

The first two tend to be more closely linked; for example, studies in the wild often suggest a specific animal model that can be tested under laboratory conditions. Typically, many laboratory studies have a tighter focus, often with specific metrics (that are believed to relate to components of a behavioural model). There is a suggestion that animal husbandry is based upon more informal observational studies. In the specific case of the otter, there seem to be very entrenched views: "The only way to keep otters in captivity is [my way]"

The real issue is how to link these various behavioural data items and combine them with other measures (such as physiological measures, reproductive success and pathology) to provide not only an integrated data-source (essentially a corpus of knowledge about the otter), but one that can be mapped on to a comprehensive model (with an emphasis upon the husbandry and welfare outcomes). This is not a trivial problem and goes along with the development of suitable models and metrics. These latter have not only to be acceptable to the scientific community, but also have to be acceptable and usable by the "practitioner" community

USER INTERACTIONS

Three kinds of user interaction are envisaged:
  1. addition of data to the data-corpus;
  2. updating of data in the corpus;
  3. querying.
The diagram below (Figure 2) illustrates this:

User Interactions with KASBAH System
Figure 2: User Interactions with KASBAH

The input to the system, in the form of new observations, or updating of extant entries, will be Web forms, but may consist text or reference to papers in journals. This cannot be determined yet. It may prove necessary to have a human moderator to convert data supplied into a form suitable for input into the KASBAH system, or it may be possible to completely automate this process.

The output from the system will be in response to queries from keepers (typically) received as Web forms. The output will not only contain 'Answers to Questions' but an indication of the reasoning.

An example question that a keeper might have is:

"In the wild, Amblonyx cinereus rootles in soft mud in river banks, estuaries and so on. Because of ease of maintenance we keep them on pebbles with no access to mud. Will this compromise the animals' welfare?"

The way this question would be presented to KASBAH would be via a set of Web forms, whose exact format cannot be determined yet, but might look something like Figure 3.

Sample Query Form
FIGURE 3: Sample Query Form

The output from KASBAH might then look something like Figure 4 below; again, it is not possible to be exact, but this is a kind of guideline.

Sample Output Page
FIGURE 4: Sample Output from KASBAH

DELIVERABLES

The major immediate deliverable from a successful KASBAH project would be a usable information system about the otter with a focus upon their husbandry and welfare in captivity. The system would also be a useful resource for animal behaviourists working with the otter: providing an integrative framework into which new information can be slotted. The provision of a Web front-end for delivery should maximise usability of the underlying system.

Although the pilot KASBAH system will use the otter (and specifically the Asian Short-Clawed Otter) as the subject animal, the overall approach and architecture of the underlying IT system will be such that it can be adapted to other species, chiefly by the addition of data (although there may well need to be specific work on the development of a suitable integrative model and associated schema and metrics).

MOTIVATION

MY MOTIVATION

This project would combine my twenty years' of IT experience in a leading-edge laboratory with my long-standing interest in Natural History and Animal Welfare. (I am a member of the Mammal Society and the Natural History Museum and a Fellow of the Zoological Society of London, and support the RSPCA, IFAW and Blue Cross.) I also belong to the Rutherford Environmental Group, a staff association at the Laboratory., and I maintain an otter web site, Otter Joy

. Although over time, my interest has covered most animal groups (with the exception of primates), I am particularly interested in the Mustelidae, especially the Lutrinae. Visiting the specialist otter collections in this country, talking to the staff and owners, and superficially comparing the conditions with those in the many zoos that also keep otters (usually A. cinereus) is what led to the idea of KASBAH.

In my opinion, it is work that needs to be done. There is a clear need voiced by some of the practitioners to quantify and objectify otter welfare. I have an interest in the species and the appropriate IT background required: hence I formulated this proposal.

CLRC'S MOTIVATION

The whole flavour of the work is multi-disciplinary with IT supporting and integrating knowledge from a variety of bio-medical areas. Government and Research Council policy is pushing both for multi-disciplinary work and for the use of computing support in all applications area (the GRIDs funding as an example of this). Given the nature of the work, it might well be possible to interest one or more of the funding research councils (e.g. NERC, BBSRC, MRC) in the utility of this activity beyond the initial application area. There would also specific benefits for my department (Information Technology Department at Rutherford Appleton Laboratory).

  1. The work is in a bio-medical area (which both ITD and CLRC see as a potential growth area).
  2. It involves the development of novel methodologies and techniques (data integration, analysis, data mining and inference; Web site and interface design) that will be of use in a number of application areas.
  3. If built generally enough, it could be extended to other applications e.g. customer management, or selling the idea of GRIDS to medical personnel. Other potential application areas - bio-medical eg protein structures, pathology, other animals, patient management and so forth.
  4. Producing something that could be used/developed/sold for other applications.
  5. It is a software area in which we are interested, and which would contribute to the development of the ITD skills base in leading edge technologies (in part by the development of my own skills and experience).
  6. The immediate application area of the system is one in which I am very interested. This would be a considerable motivator for me to gain new skills of use to the Department.

COMMITMENT

CLRC'S COMMITMENT

ITD has agreed to pay course fees, allow limited business visit time (to be negotiated ad hoc with line managers), allow up to 10% work time per week to be used on the project (unless required for immediately work-related training, as agreed with my line managers), full use of professional library services, server space and use of my work PC for the project. In addition, I may freely approach IT specialists in the Department for advice on various aspects of novel techniques.

MY COMMITMENT

I am prepared to commit most of my "free" time to this project. Since I work four days per week for CLRC, this would translate into roughly 2.5 days per week, including the Department's time agreement . I would pay non-fee costs such as travel, accommodation, meals, books and materials, and use my own PC for the work in that time.

CONTACT INFORMATION
Miss Lesley C. Wright Tel: 01235-44-6563
R1, 1.58 Email: L.Wright@rl.ac.uk
Rutherford Appleton Laboratory URL: Otter Joy
Chilton, Didcot
Oxon,
OX11 8SR

BIBLIOGRAPHY

These are some of the references I have used in developing this proposal. It is by no means exhaustive.

CHANIN, P (1985). - The Natural History of Otters. Croom Helm Ltd, ISBN 0 7099 3401 7

COOPER, J.J. & MASON, G.J. (1997). - The Behavioural Priorities of Mink (Mustela vison) in a Closed Economy. Proc. of the British Soc. of Anim. Sci. : 17

DUPLAIX, N (1980). - Observations of the Ecology and Behaviour of the Giant River Otter Pteronura brasiliensis in Suriname. Rev. Ecol (Terre Vie), vol. 34:495-620

HAWKE, L., LAUER, D.B., & STEIN, Z. - Effects of Increased Food Dispersal and Random Feeding Time/Place on Stereotyped Behaviours of Otters at Adelaide Zoo. http://www.zoonews.ws/IZN/299/otters.html

HARRIS, C.J. (1968). - Otters, A Study of the Recent Lutrinae. Weidenfeld & Nicolson, London.

KRUUK, H. (1995). - Wild Otters : Predation and Populations. Oxford University Press, ISBN 0 1985 4070 1

MASON, G.M. (1993). - Age and Context Affect the Stereotypies of Caged Mink. Behaviour, 127: 191-229

MASON, G.M. (1991). - Stereotypies and Suffering. Behavioural Processes, 25: 103-115

MENDL, M. (2001). - Assessing the Welfare State. Nature, 410: 31-33

MENDL, M. (1997). - Responses to Supernormal Stimuli and their Implications for Animal Welfare: can animals have too much of a good thing? Applied Animal Behaviour Science , 54: 47-51

MENDL, M. (1991). - Some Problems with the Concept of a Cut-Off Point for Determining when an Animal's Welfare is at Risk. Applied Animal Behaviour Science, 31: 139-146

SIVASOTHI, N. (1998). - Asian Otters in Captivity. Asian Otter Newsletter, 5(1): 15-18

WEBSTER, A.J.F. (1998). - What Use is Science to Animal Welfare? Natur Wissenschaften, 85: 262-269