| CITEX
|
|
|
| Introduction and highlights on each function of CITEX
|
CITEX currently provides the following services:
|
| Assessment of a Farm: The Assessment Subsystem
|
This subsystem has two functions. The first one is to evaluate a new farm in a given location for determining the possibility of cultivating citrus in it (a feasibility function).
The second function is to evaluate a given farm that is already cultivating citrus in terms of its productivity and to
give recommendations concerning how to improve its conditions to get more yield.
|
| Irrigation Scheduling: The Irrigation Subsystem
|
The main goal of this subsystem is to produce a schedule for irrigation
of a particular farm. An irrigation schedule should demonstrate the water
quantity related to each time instance. The calculated water quantity should
also be adapted according to some farm characteristics such as intensity
of plants, efficiency of drainage system, etc.
|
| Fertilization Scheduling: The Fertilization Subsystem
|
The main goal of the fertilization subsystem is to determine the fertilization
requirements for the citrus crop. Fertilization requirements include several
aspects such as the fertilizer's type, quantity, the fertilizer's application method, and
the intervals between applications.
|
| Disorder Diagnosis: The Diagnosis subsystem
|
This subsystem performs two primary functions. First, it concludes the
causes of user complaint. alternatively, it can verify/reject a user assumption if
the user suspects a given disorder(s) and wants to obtain a treatment. This subsystem
takes into account the possibility of a plant being infested with more than one
disorder at a time, and so the user can select more than one value for an
attribute of an observation. The system can conclude the causes of all user
complaints in one session.
|
| Disorder treatment: The treatment Subsystem
|
| The purpose of this subsystem is to provide the user with
appropriate treatment for the infected plants.
Back to the beginning of this page
|
| Platforms and Methodology |
| Hardware:
PCs with the following H/W specification are supported:-
- 80486 M/C.
- 8 MB RAM.
- 10 MB of HD should always be free for swapping.
- CPU in real mode (does not work under windows).
Software:-
The following software is required:-
- MS-DOS 3.1 or higher.
- MS-Dos Arabization S/W or its equivalent.
|
| Methodology |
KADS
The methodology followed in the construction of CITEX is KADS. According to the KADS approach, the construction of KBSs
goes through a fixed set of stages. Each of these stages results in a particular
model, i.e., the KADS view of knowledge modeling requires the use of multiple
models. Each model is used to handle different aspects of the knowledge base. The idea behind the multiple model principle of KADS
is that the knowledge engineering space of choices and tools can, to some
extent, be controlled by the introduction of a number of models. A model
reflects, through abstraction of detail, selected characteristics of the
empirical system in the real world. These models are: the
organization model (which provides an analysis of the socio-organizational
environment in which the KBS will have to function), the application model
(which defines: the problems that the system should solve within the organization,
the functions of the KBS in the organization, and the
external constraints that are relevant to the development of the application),
the task model (which specifies how the system functions as specified
in the application model, in terms of a number of tasks),
model of cooperation (which contains a specification
of the functionality of the sub-tasks in the task model that require
a cooperative effort), the model of expertise (which specifies the problem
solving expertise required to perform the tasks assigned), and the design model
(whereby, the description of the computational
and representational techniques for the system are specified). To model
the expertise, the KADS approach assumes that it is possible and useful to
distinguish between several generic types of knowledge according to the
different roles that the knowledge can play in the reasoning process. Also,
the KADS approach assumes that these types of knowledge can be organized
in several layers, which have only limited interaction. The categories
in which the expertise knowledge can be analyzed and described are based
on epistemological distinctions: they contain different types of knowledge.
The distinction is as follows:
- Domain knowledge: static knowledge describing a declarative theory of the application
domain.
- Inference knowledge: knowledge of different types of inferences that can be made in this
theory.
- Task knowledge: knowledge representing the elementary tasks.
- Strategic knowledge: knowledge which controls the overall reasoning process.
Typologies of elements of a model of expertise, such as a typology of
knowledge sources, represent a first step into the direction of reusability.
A further step would be to supply partial models of expertise such as models
without all the detailed domain knowledge filled in. Such partial models
can be used by the knowledge engineer as a template for a new domain and
thus support top-down knowledge acquisition. In KADS such models are called
interpretation models. The KADS interpretation models are models of expertise
with an empty domain layer. Interpretation models describe typical inference
knowledge and task knowledge for a particular task.
Back to the beginning of this page
|
| Validation results |
| In each subsystem, a performance factor is calculated for each of the
best three human experts and the CITEX expert system. This factor is calculated
as follows:
Gi = Pi / No. of cases
Where "Gi" refers to the average score of expert number "i" and "Pi"
refers to the score of expert number "i". table 1. shows the calculated
performance factors. Expert 1, in this table, is the best domain expert
performer, expert 2 and expert 3 are the second and third best performers respectively.
Experts 1,2,and 3 are different human domain
experts for the different subsystems.
table 1. Performance Factors
|
Subsystem
|
CITEX
|
Expert 1
|
Expert 2
|
Expert 3
|
|
Assessment
|
3.5
|
3.4
|
3.3
|
3.05
|
|
Irrigation
|
2.86
|
2.9
|
2.78
|
2.68
|
|
Fertilization
|
2.88
|
2.67
|
2.5
|
2.18
|
|
Diagnosis
|
3.54
|
3.13
|
2.81
|
2.71
|
|
treatment
|
3.37
|
2.83
|
2.81
|
2.74
|
|
Overall
|
3.16
|
2.93
|
2.79
|
2.51
|
This table shows that in all subsystems, except for the irrigation subsystem,
CITEX is the best performer and it has the best overall system performance.
Back to the beginning of this page
|
| Developers and Domain Experts |
| Developers
Domain Experts
- Prof. Kamal El_dien Abd_Allah
- Prof. Ahmed Efat
- Prof. Ahmed Abdel_Fatah
- Prof. Badwy Abdel_Rahim
- Prof. Ahmed Abdel_Aziz
- Prof. Salah El_Eraky
Back to the beginning of this page
|
| list of all versions |
- An expert system was developed to assess the feasibility of citrus cultivation (CITEX/ Feasibility). For the present, CITEX can only be applied to oranges, which represent approximately 75% of citrus production in Egypt. This expert system supports decisions regarding whether or not a unit of land is suitable for cultivation, June 1991.
- The ClTEX/agro-management expert system's objective is to generate an irrigation, fertilization and crop management operations schedule. Irrigation and fertilization operations include water quantity, irrigation interval, nutrient quantity and application interval. Preventive operations include agriculture practices and
preventive chemical spraying, Dec 1993.
- ClTEX/Disorder Handling identifies the cause and severity of 50 observable disorders and then proposes a course of treatment, Dec 1993.
- CITEX production version 1.0 implemented in KROL has been issued in Dec.94.
- CITEX production version 2.0 which is a bilingual version (Arabic/English) has been issued Jan. 95.
- CITEX production version 2.1 including images in the treatment subsystem has been issued Feb. 96.
- CITEX production version 2.11 enhancement version of 2.1 has been issued May. 96.
Back to the beginning of this page
|
| Sites where system is deployed |
|
Citex has been distributed to: four horticulture sites, two faculties of agriculture, two research institutes and five private sector farmers. Back to the beginning of this page
|