|
*** Polytechnic Higher National School of Yaounde (P. H. N. S)
Department of Computer Science and Mathematics (D. C. S. M) *** Laboratory of Data-processing, Multi-media and Applications
(LABODAMA) |
END OF TRAINING THESIS TOPIC : « KNOWLEDGE DISCOVERY IN A
MULTI-MEDIA DATA BASE AND MULTI-POINT OF SIGHTS : APPLICATION TO MEDICAL
RESEARCH » PRESENTED AND SUPPORTED IN YAOUNDE ON JUNE 19, 2000 BY MFEGUE GUY PHILIPPE FOR OBTAINING THE DIPLOMA OF COMPUTER SCIENCE AND
DATA-PROCESSING ENGINEER |
Thesis Document *** (c)
GPM, 2000 |
||
|
|
||||
|
THE JURY |
q
PRESIDENT : |
ü
Dr CLAUDE TANGHA – SENIOR LECTURER IN THE D.C.S.M OF THE P. H. N. S & RESPONSIBLE OF THE
LABODAMA |
||
|
q
MEMBERS : |
ü
Dr PAULINE LAURE FOTSO – SENIOR LECTURER AT THE FACULTY OF SCIENCES OF THE UYI & THESIS DIRECTOR ü
Dr CANAL – MULTI-AGENTS SYSTEMS THEORY SPECIALIST & GUEST MEMBER ü
Dr GUILLAUME KOUM DISSAKE – SENIOR LECTURER IN THE D.C.S.M
OF THE P. H. N. S ü
Dr AUGUSTIN YEKEL - SENIOR LECTURER IN THE D.C.S.M
OF THE P. H. N. S |
|||
|
THE CANDIDATE |
|
MFEGUE GUY PHILIPPE |
||
|
|
|
|||
|
ABSTRACT |
The conservation of the data on
data-processing supports makes it possible to file the cognitive inheritance
which one can then consult constantly to get information and discover new
knowledge rising from varied associations of the existing data. The
extraction of knowledge resulting from the data requires techniques such as the
Data Mining and the Logic. This discovery of knowledge can involve a multipoint
of sights analysis because of diversity of the experts and complexity of the
fields. The present thesis is devoted to the
design and the realization of a system enabling knowledge discovery in a
multi-media database (which integrates pictures, text, sound and video)
comprising various points of view. To put this approach into practice,
we used as basis the MEDITRA Project initiated in the Faculty of Science of
the University of Yaounde I in 1995. The focus point of that project is to
collect and store knowledge on African traditional medicine, in order to
place them at the disposal of the researchers and experts of the medical and
pharmaceutical fields for exploitation. For that purpose, we have :
firstly explored the work already done within the framework of the MEDITRA Project.
Secondly, we modelled the system Meditra
Software Manager to be carried out. This system must manage three
databases in interaction: the database of the medicinal plants, that of the
drugs and that of the diseases. Since the existing comprised only the first
two databases, we had to model and implement the third database. Thirdly we
worked out an original approach by fetching inspiration into the Multi-points
of Sights Technique developed by the theory of Agents and Multi-Agents Systems
according to MASSAGE 2. This approach proposes an architecture of the system,
made up, in its general aspect, of three specialized layers and, in its
functional aspect, of sights known as multiple, offering various perceptions of the same system. Through the
interrogation of the mining and databases, results and new knowledge are
provided by the system, making it possible to meet various needs. Key
words
: Knowledge Representation, Data Mining, Multi-media, Databases, knowledge of
Authorware. |
|||
|
|
|
|||
|
GENERAL INTRODUCTION |
In a World where many diseases still constitute
public health problems, the discovery of new therapies, the invention of
drugs and top vaccines are the only weapons able to make it possible for mankind
to take up this medical challenge. Since the active ingredients of the drugs
derive from the plants and the chemical compounds, medical research rest mainly
nowadays on the plants and polymers resulting from chemical laboratories. In Cameroon and World-wide, the Medical
Research pay particular attention to the plants known as medicinal and to the
local pharmacopeia. However, at a given time, there was a problem of
coordination of the research tasks, which started to find ways of solution
through projects of follow-up and rationalization of medical research such as
the MEDITRA project initiated at the Faculty of Science of the University of
Yaounde I in 1995. Works consequently were undertaken in this groundwork like
those having made it possible to produce two applications called SYGPLAM
(Medicinal Plants Management System) and POSOFT (Potions Software); The first
application manages a medicinal plants database whereas the second ensures
the management of the drugs resulting from these plants. For a better coordination
of the efforts of research and an optimization of the results already
obtained, the need to centralize all these assets was felt, thus constituting
the object of our work. Indeed, our contribution to the MEDITRA
project consists in designing and carrying out a system enabling knowledge discovery
in a multi-media database using the Multipoint of Sights approach by the
means of the Multi-agents. In other words, we bring our stone to improving, completing
and federating the above-mentioned systems and applications related to the medical
research using the techniques of the Artificial Intelligence coupled to those
of the Multi-media Databases. The work presented in this thesis is organized in
five (05) chapters. Chapter I presents the existing. Chapter II is
devoted to the review of the concepts essential to our topic and which are
Representation of Knowledge, Data Mining, Multi-media Databases, Multi-agents
Systems and Multipoint of Sights. The third chapter allows, under the light
of the preceding concepts, to circumscribe, in its first part, the problems
to which the awaited results will be used as solution; the second and third
parts of the same chapter will relate to the conceptual analysis and the
modeling of the system to be realized, which includes the application named SYGMA
(Diseases Management System , one of the subsystems of the system to be
realized, in charge to manage the diseases). In chapter IV, we will consider
a proposal for an architecture. The chapter V will relate to the putting into
practice. By concluding this work, we will certainly recall the problem
arising, the stages of study and the stage of realization to which we
arrived, but also we will forecast axes of continuation and finalization of
the system conceived and partially carried out, and we will make proposals
for an orientation of medical research in order to make it benefit from the
opportunities offered by the unceasingly influence that Computer Science is exerting
on many human activity branches. |
|||
