THE FUNCTIONALLY-ORIENTED METHOD FOR SPECIALIZED EI-SYSTEMS DESIGN APPLICATION

. The algorithm and the model of the functionally-oriented method for specialized educational and intellectual systems (EI systems) automated design, based on Petri nets, are developed in the article. The example of the developed algorithm application for the real EI-system design is also given. The developed model based on Petri nets helps to identify the major states of the system synthesis process.


Introduction
Development of any complex system is associated with some major difficulties.The initial design stage requires determination of the system structure, its main elements and parameters, relations among data structure elements and others.There are number of methods for the design automation at the system level [1,5,9,14,16], but one of the most effective for specialized EI systems development is the functionally-oriented method [11,15].
Examples of such complex systems are specialized educational and intelligent systems (EI-systems), which are nowadays very popular and fast-growing.Since EI systems ensure effective support of modern training technologies [6], development of the model and algorithm for functionally-oriented method application is the relevant modern task.

Development of the specialized EI systems automation design algorithm
This article presents the algorithm for the functionallyoriented method application, which involves the following basic steps: Step 1. Input data setting and defining the EI systems class.
Step 2. Selection of the EI-system functional profile [8].
Step 3. Formation of the projects that correspond to the selected (developed) functional profiles of the developed EI-system.
Step 4. Specifying of the linguistic variable "EI-system project general coefficient".
Step 5. Selection of the partial coefficients for all EI-system projects.
Step 6. Determination of the estimates for all EI-system projects partial coefficients.
Step 7. Determination of the estimate for each EI-system projects general coefficient.
Step 8. Comparing EI system projects by general coefficients and selection of the technical implementation project.
Step 9.If the project satisfies customer requirements, then the algorithm is completed, otherwise, go to step 2.
According to the developed algorithm, one or several standard functional profiles should be selected (step 2 of the algorithm) from among those, which correspond to the class specific for EI-system, which are designed according to the technical task.It is possible to develop a special functional profile or modify the standard one.During the linguistic variable (LV) "EI-system project general coefficient" introduction (step 4) all terms and the corresponding functions have to be built.When choosing partial coefficients for all EI system projects (step 5) the appropriate LV should be set and the priority coefficients should be calculated.Evaluation of EI-system projects partial coefficients (step 6) is determined by expert evaluation, calculation and analysis based on the unified software modules values that are part of this project.
The developed algorithm allows automatically implement the computer aided design method and realize the specialized EI-systems synthesis procedure.

The EI systems designing process model based on Petri nets
For studying of the basic states of functionally-oriented method based on the developed algorithm, the model basing on Petri net [7,10,13] was built, it can be described using the following expression, an example of the scheme model is in Fig. 1.
where:  The marker in this position demonstrates that the project satisfies the customer requirements.

 
p9 Position responsible for the case, when this project does not meet the customer requirements.The marker in this position testifies choice of some other (different from the previous) EI-system functional profile and subsequent formation of the next project, which would satisfy the requirements of the customer.-This copy is for personal use only -distribution prohibited. - The built model based on Petri nets contains number of states р1-р10, and number of transitions t1-t8, which have individual logically-functional purpose (see table 1

and table 2, respectively).
The positions set describes the possible model`s states and enables tracking behavior of the design process on separate areas, whereas the transitions set reflects the internal processes that occur inside the model and are designed to achieve results.
Below, in fig.2, the corresponding states reachability graph for the developed model based on Petri net is presented [12], it allows exploring possible states and dynamics of the specialized EI systems design process.The subsystem configuration setting. t2 The developed EI-system class determination. t3 The developed EI-system functional profile choice. t4 Formation of projects that correspond to the selected (developed) functional profiles of the developed EIsystem. t5 Choice of the partial coefficients for all EI systems projects.Determination estimates of the partial coefficients for all EI systems projects.

Results of the specialized EI system structure synthesis using the functionally-oriented method
To illustrate the developed method for EI system design, the following example is presented.Assume, that the EI system, which according to future plans and other considerations is referred to class "12.Context-search EI system", is being developed.Let assume, that for this class of systems there are two standard functional profiles: (3) (since this example illustrates design procedure for only one EI-system, p index in the project conditional signs is excessive, so here and further on we will be limited to only  index use).
For the optimal project selection the partial modules coefficients listed in table 3 (in table 3 and further on the letter t after shortenings VHV and HV should be read as "trustworthy") should be considered.Some partial coefficients values are presented in conventional units, the rest, primarily integrated indices are characterized by fuzzy expert estimates [2,3,4] for LV with five terms and changes range from 0 to 1. Limits of the according belonging functions are defined in analogy with the common criteria.
 projects the optimal one should be chosen, with considering that the most important system The cost of the project will be considered to be equal to the sum of the included modules cost.For those projects, where only one module identifies some indicator, its value for the entire project coincides with one for indicator`s module.For other indicators their scores for the entire project also correspond to the LV in the range from 0 to 1, and for their determination the same methodology as for general project index finding should be applied.In this case, priority coefficients help to verify that various modules contribute differently to this index value.


Search speed HV T 0.9 .
Search precision HV T 0.9 and VHV T 0.1_.

Reliability
For example, Вi coefficients will be got with the use f Fishburn method: where lknumber of k-th module in the order of its importance in the project`s appropriate metric, T is the total number of project modules.
Let the module 4  be more important for the 8 w coefficient, than the module 45  but module 45  is more important, than 41  one.According to formula (5) the corresponding modules priority coefficients will be computed: Then the following data will be got for the project The estimation of the  In the same way the reliability indicators for four other projects will be found.For the project   ) the last will be computed based on the data summarized in table 5. Guarantability indicators for the remaining three projects can be found in the same way.For the project Results of the belonging functions for each 1  project indicator calculations are given in table 7. -This copy is for personal use only -distribution prohibited. p8

p10
Model shutdown position.The marker in this position testifies completion of the developed model work.T h i s c o p y i s f o r p e r s o n a l u s e o n l y -d i s t r i b u t i o n p r o h i b i t e d .

t6
Comparison of EI system projects by general coefficients and selection of the technical implementation project.t7 Start of the other functional solution selection mechanism for project implementation that ensures the customer requirements.t8 Implementation of the project.Shutdown of the model.

where 7  16 
-"Contextual search" functional service; 19  - "Ranking of search feedback by relevance" functional service; 27  -"Automated text summarization" functional service; 31  -"Automated text annotation" functional service.Assume that system developers have access to the standardized software modules library, which among others contains the following modules: is "Automatic correction of the search query spelling" functional service.Thus, in accordance with the proposed methodology for the developed system implementation the 4 following projects might be formed:

9  1 w 6 w
T h i s c o p y i s f o r p e r s o n a l u s e o n l y -d i s t r i b u t i o n p r o h i b i t e d .for the respective projects.The projects comparison will be done with the thought that the general quality indicator for each of them will be determined based on the following partial indicators (presented by decreasing priority): search speed (cumulative characteristic determined by the modules reliability and quality of their license maintenance); cost of the project 7 w .

 1 Lets find the guarantebility index value 6 w 9 w
Summarizing, annotation quality VHV T 0.8 .ReliabilityAV T 0.9 and LV T 0.1 .Cost 1800 $.License terms LV T .First of all we have to determine the reliability f the each project (8w ) and quality of the license maintenance by the modules developers ( ).For each term of the same content (VLV, LV, AV, HV, VHV) we will get the estimates:

1  project generalized reliability index 8 w
will be done by the following formula:

4 
(0.633) = 1 and for all other terms with the belonging function equals zero.

2 
the T h i s c o p y i s f o r p e r s o n a l u s e o n l y -d i s t r i b u t i o n p r o h i b i t e d .

Table 5 . 4 
Data for the first project guarantability index calculation (0.6) = 1 and to all other terms with the function equals zero.

Table 7 .
The first project partial indicators belonging functions T h i s c o p y i s f o r p e r s o n a l u s e o n l y -d i s t r i b u t i o n p r o h i b i t e d .

Table 2 .
The developed model based on Petri net transitions table

Table 3 .
Parameters of the unified software modules

Table 4 .
Data for the first project reliability index calculation