Systems analysis: methodological and meaningful aspects.

Systems analysis is one of the youngest methodological tools that became a relatively independent method in its own right in the mid-1900s. Currently the theory and practice of systems analysis continue their rapid development. Different views are expressed as regards its content and its relationship with systems approach, operations research and decision theory. This is why, despite the existence of a considerable number of works devoted to systems analysis, a need is felt for some generalization and specification of its provisions. It is also necessary to consider systems analysis as a full cycle of research involving different methodological approaches and methods. This is all the more important in a situation where the goal programming method based on systems analysis procedures is used on a broad scale in the armed forces organizational development planning.

Even though systems ideas sprung up approximately in 2,500-2,000 years B.C., the systems movement came into its own only after the Austrian biologist and philosopher Ludwig von Bertalanffy (1901-1972) developed the general systems theory in the late 1940s. In 1954, the Society for the Advancement of General Systems Research was founded. Currently the scientific community recognizes the existence and development of the systems science. It is a vast complex of scientific disciplines and scientific trends of different nature and includes the general system theory and specialized, particular systems theories. (1)

The general systems theory is a general system theory (metatheory) of systems. It contains the most general system tenets of importance for systems of any nature and performs the methodological function in respect of specialized system theories. In this connection, the general system theory is also called the logical methodological metatheory.

There are many specialized systems theories. These are systemology, cybernetics, informatics, synergy, systems approach, systems analysis, operations research, decision theory, systems technique, and others. Among these, of particular importance to military researchers are systems approach, systems analysis, operations research and decision theory.

Systems approach is an important methodological trend in scientific investigations and social practice. It is based primarily on philosophical premises as applied to the study of object systems, general systems theory and a number of particular systems theories.

The initial philosophical premise underlying systems approach is the system principle of reality. This principle is closely linked with such crucial principles of dialectics as the principle of universal connection and interaction and the principle of qualitative difference of part and whole. But as a specialized general scientific methodological concept, systems approach is in a naturally subordinated relation to dialectics: it is a concrete definition of the principles of dialectics as applied to research, designing and development of object systems.

In the most general sense, the main feature of systems approach, one that distinguished it as an independent methodological concept, consists in using the fundamental methodological notion of system as an abstract unified image of investigated concrete objects of any nature; in singling out the leading, determining aspects and tendencies in a system's development; and in representing knowledge and a problem or task tackled as a conceptual system of sorts. Its high degree of generalization is something to be noted in particular. This is due to the system nature of reaity. By virtue of this circumstance, each researcher or practical worker should take into account, in all spheres of his activities, the provisions, requirements, principles and aspects of systems approach.

At the same time, any approach in theory and practice is still an approach. It defines a researcher's or a practical worker's viewpoint of principle on a subject under consideration and the general landmarks in cognizing the reality. But to deal with concrete scientific or practical problems, a researcher or practical worker is due to investigate within the framework of systems approach a complex of concrete methodological assets that make it possible to address these problems correctly.

This complex of methodological assets used to validate solutions to complicated problems in most different spheres of human and social activities was given the name of systems analysis. Analysis and synthesis are organically combined in the systems analysis process. This characterizes the decision validation procedure that consists in decomposing a problem into components that more easily lend themselves to study, in using the most appropriate methods to deal with particular tasks, and, finally, in generalizing particular conclusions and solutions.

The main principles of systems analysis are the following (2):

* sense of purpose (the main goal should be determined and clearly formulated at early stages);

* subordination of particular goals to the main one;

* consideration of objects of study or problems tackled as systems;

* identification and analysis of several variants of a solution to a problem and the choosing of a rational or efficient variant;

* complex and balanced nature of research methods applied;

* leading role of decision-maker (DM) at main stages in the process of validation of solutions to difficult problems;

* organization, methodological support and pursuit of systems research by the systems analyst, the central figure in the systems analysis procedure;

* first person principle--while preparing, validating and participating in realization of a solution, the systems analyst must have an opportunity of constant communication with the DM or his competent representatives;

* the systems analyst shall have access to all information related to a problem tackled.

It is very important for researchers to know primarily the sequence of works performed while systems analysis procedures are applied, i.e., to know the main stages in the systems analysis. At the same time, specialists are not of a single mind on this matter. Different sources cover different variants of systems analysis, doing so either in an overly general way or in relation to concrete problems, each of which is characterized by systems analysis specifics of its own. In this connection it makes sense, while describing stages of systems analysis, to engage in systems generalization. In so doing, we will abstract ourselves from particular specifics of problems and single out the main stages that are general for most types of problem situations and their execution sequence.

Besides, it is currently impossible to have just one single exposition of even the most important stages in systems analysis. The thing is that two trends in systems analysis have become current at the present time, and a number of stages in these are substantially different from each other. The first trend is about analyzing difficult problem situations with the use of models that make it possible to determine attractiveness indicators of alternatives without modeling systems' operations. The second trend is based on efficiency theory methods which envisage the use of models of complex systems operations. Figure 1 shows the main stages in systems analysis for those two trends.

The first two stages are practically the same for both trends. The third stages have differences in names, but they are near in content. The first trend envisages validation and choice of criteria which will help estimate alternatives' attractiveness indicators from the point of view of their conformity to the aim. A vector of target indication parameters is formed for the second trend. It is also a list of criteria but it established evaluations for criteria that are required from the point of view of attainment of the main aim. After the criteria or the target indication parameters vector is validated, the admissible alternatives (strategies) of a solution to the problem in hand are developed, to wit, the alternatives that secure solution to the problem and are rational from the point of view of aim attainment.

The main feature of the systems analysis variants for the first trend consists in the importance that attaches to the making of a model to determine indicators of attractiveness alternatives. The model reflects a system of links between components of a problem and factors influencing attractiveness of alternatives. The second trend envisages validation of decisions involving the use of efficiency theory methods. Figuring importantly in its context are efforts to create a dynamic complex system operation model.

[FIGURE 1 OMITTED]

The seventh and subsequent stages in systems analysis are identical for both trends. They are executed with the use of methods of the multi-criteria theory of decision-making. In this context, the decisive rule is the first to be developed that serves to compare the alternatives. After that the alternatives are compared and one or several best ones are chosen on the basis of alternatives' attractiveness indicators or operations efficiency indicators. But it is the DM who has the prerogative to make the final choice of the best alternative on the basis of his inexplicit system of preferences.

The tenth stage of systems analysis is of much importance for solving problems in a complex organizational-technical system. This stage implies participation of systems analysts in the implementation of an accepted decision and therefore its other name is decision implementation control. It is due to the fact that no preliminary check of good quality of validated decisions is possible for problems where decisions are validated with the help of systems analysis. In this connection, systems analysis procedures are, as a rule, of iterative nature: they can rarely be completed during one full cycle. Usually systems analysis involves repeated reiteration of full-format investigations or throwbacks to investigations that start at one of intermediate stages. Fig. 1 shows this with reverse arrows, including for cases where the DM is dissatisfied with a solution or a problem has not been removed.

Singling out stages in modern systems analysis at the general scientific level is not an aim in itself. As is evident from experience, complex investigations into military problems are, as a rule, interdisciplinary. They use provisions, approaches and methods of both military science and other components of the system of sciences. Moreover, to validate decisions on many problems, it is quite enough to perform the main stages of systems analysis without introducing additional stages and sub-stages.

Two interconnected parts can be singled out in the systems analysis procedure for both trends. The first of these includes stages from one to five, and their execution is of creative nature and has no cut-and-dried rules and algorithms. Heuristic and logical-heuristic methods have a big role to play in this sense as do concrete modeling methods at stage five. Thus, a problem situation model results from part one of systems analysis. Part two includes all other stages that are connected primarily with choosing from admissible alternatives with the use of methods of multi-criteria decision-making theory. The totality of the two parts of systems analysis forms a decision validation model with regard to a problem under consideration.

Thus, while the systems approach is a totality of ideas and principles of systems research, systems analysis is its prescription methodological realization where decisions on complex problems are validated. Such is the present-day relation of systems approach and systems analysis. Notice that they were often identified with each other at the early stage in the systems movement. One dictionary, for example, says that "in a broad sense the term 'systems analysis' is occasionally used as a synonym of systems approach." (3) This interpretation was obviously due to an insufficient development of systems analysis provisions, primarily the lack of well-substantiated methods of multi-criteria decision theory. Currently these methods are in existence, and so systems approach and systems analysis have been given qualitative definiteness as particular systems theories.

Different variants of the general systems analysis structure can be used on the basis of the just considered two trends during investigations. Five main variants may be singled out among these.

The first two variants are in direct conformity with the above two trends. This means they envisage the singling out of the best alternative when problem situation models and decision theory methods are used. They are also employed when the main aim is sufficiently clear and on top of it is formalized for the second trend.

The specific feature of the third variant of the general systems analysis structure, one based on its second trend, consists in using a combination of different operations models of systems under consideration (for example, analytical and imitation) in order to enhance the degree of validation of decisions produced.

The fourth variant envisages integration of the first and second trends in systems analysis. It is used to prepare, for example, operations to be pursued by force groupings. In this context, the first trend is used to choose a rational variant of a grouping's combat strength level, primarily the numerical and qualitative composition of weapons and military equipment. To determine alternatives' quality indicators, the model of a problem situation is drawn--as a rule, on the basis of methodological provisions of qualimetry. After a systems analysis procedure is accomplished with regard to the first trend, the same procedure is performed for the second trend. It enables determination of rational methods of operations by a force grouping. Integration of the two systems analysis trends can considerably enhance validation of decisions and plans accepted.

The fifth variant of general structure of systems analysis procedures can be organized for complex problem situations. It is the most characteristic one where validation of strategic-level decisions is involved; specialists call it planning. (4) These levels are characterized by a considerable vagueness of the main aim. But if an aim is determined wrongly, a rational decision will not be accepted, with all the ensuing negative consequences. In this connection, two interlinked first-variant systems analysis procedures are performed in the process of planning. The first of these is intended to validate the choice of the main aim, while the second that of rational ways and methods of its attainment.

The just considered five variants of systems analysis are decision validation models for complex problems or systems analysis models for research. In all evidence, specialists from different areas of science and practical work can come up with other variants of systems analysis procedures.

Apart from systems analysis, operations research is concerned with validation of decision-making. In this connection, one should be clear about things that are common to systems analysis and operations research, as well as differences between them.

As is common knowledge, operations research is an applied scientific discipline that develops and employs quantitative methods to validate decisions. The term "operations research" sprang up during World War II, when an urgent necessity emerged of drawing up proposals on how to deal with a number of difficult problems for the heads of the armed forces (primarily British and U.S.). Specialized combined teams that included scientists of different professions (mathematicians, physicists, engineers, biologists, etc.) were set up for the purpose. Initially those problems were about rational distribution of different forces and assets and arms employment. Later operations research methods came to be used on a broad scale in all spheres of society's activities.

Various works also offer many variants of stages in operations research. For example, one of the well-known foreign specialists, H. Taha, included in the operations research process the following five stages (5): identification of a problem (formulation of research aim, identification of possible alternative solutions as applied to a problem situation under consideration, determination of requirements, conditions and restraints inherent in a system under study); construction of a model; solution of a problem in hand with the help of the model; verification of the model's adequacy; realization of research results (performed by a research group in close contact with DM, his aides and representatives).

Specialists in mathematics single out six stages (6) in operations research: formulation of a problem; choice of a model; model-assisted search for solution; solution testing; organization of control; creation of a favorable regime to realize an approved solution.

A comparison of systems analysis (Fig. 1) and operations research shows that they have much in common, including informal (creative) stages. Notice that both operations research stages we considered do not indicate such an important stage as decision-making by DM. Apart from that, the first two stages in operations research should be the same as the first two stages in systems analysis. Yet these stages are much simpler for operations research in view of greater definiteness of components of problem situations.

In his work on operations research H. Raiffa offers this quotation from E. Queid that was drawn from his address to a symposium on systems analysis and decision theory: In a broad sense, any analytical research designed to help the decision-maker to choose a preferable action from numerous alternatives could be called systems analysis. And he follows it with his own judgment: It is accepted that the term "systems analysis" refers to solution analysis in very complicated and vaguely defined problems. "Operations research" is about analyzing solutions for a more restricted class of situations, when both the structure of a problem and its aims are rather well-defined. Of course, there is no clear-cut boundary between these two categories. (7)

It follows from the foregoing that in principle the specialists are not against including operations research into systems analysis as its specific supplement for validating solutions in well-defined problems. The attributes that divide them are due only to the degree of definiteness and the concomitant complexity of problems under investigation: operations research "works" well with well-structuralized problems (that lend themselves well to mathematical description), while systems analysis does the same with weakly structuralized and non-structuralized problems (that cannot be described mathematically in full or at all). Moreover, the latter makes it possible to validate only rational, effective or Edgeworth-Pareto optimal solutions. This is of fundamental importance because works often speak about optimal decision-making on weakly structuralized and non-structuralized problems (for example, most complicated non-structuralized problems in the functioning and organizational development of the armed forces). Often one can hear or read about the "most optimal" solutions. In all evidence, it's the influence of operations research specialists and the outward attractiveness of the term "optimal."

Fig. 2 shows the connection of systems analysis, operations research and decision theory.

The contiguity, in Fig. 2, of ellipses characterizing the areas where systems analysis and operations research is applied means nearness of these theories, for they are intended to validate solutions to difficult problems. The intersection of the areas of systems analysis and decision theory shows that within the framework of systems analysis procedures multi-criteria decision theory methods are used.

That the areas of application of decision theory and operations research have a common part means that decision theory, along with others, considers optimal choice (optimization) problems which use operations research methods.

Decision theory methods may be used independently to solve choice problems if the type of problem, main aim, alternatives and criteria are clear and there is a model of problem situation. Methods of this theory are also used to solve its other traditional problems, such as distribution of alternatives by classes (sorting), or regulation of alternatives by value and importance (arrangement).

[FIGURE 2 OMITTED]

The role and importance of systems analysis in the methodological arsenal of military researchers follow already from the fact that the overwhelming majority of problems in the activities of the armed forces are weakly structuralized or non-structuralized. Two points should be singled out in the use of systems analysis for solving these problems. First, there are always numerous problems in the activities of the armed forces, whose solution is possible through the direct use of systems analysis. Second, the most difficult problems are solved with the use of the goal programming method based on systems analysis procedures.

Hence the considerable role and intransient importance of systems analysis for military investigations.

Even a cursory characterization of the main stages in systems analysis variants for the two trends reveals their complexity, particularly so if we take into account that a number of its important stages are of informal nature. Modern systems analysis demonstrates a stable tendency to elaborate methodological approaches and recommendations, which make it possible to enhance validation of its informal stages. But it is already a different theme.

Evident from the characterization of the main stages and variants of systems analysis are their entire complexity and weak propensity to formalization. This is why DM, systems analyst and experts engaged in systems analysis should creatively and rationally combine science and art, as well as logical, heuristic and other research methods.

NOTES:

1. A.I. Ouyemov, Sistemnyi podkhod i obshchaya teoriya system, Mysl Publishers, Moscow, 1978.

2. F.I. Peregudov, F.P. Tarasenko, Vvedeniye v sistemnyi analiz: Uchebnoye posobiye dlya vuzov, Vysshaya shkola Publishers, Moscow, 1989; F.G. Kolomoyets, Osnovy metodologiyi nauchnykh issledovaniy. Rekomendatsiyi po provedeniyu dissertatsionnykh issledovaniy, Harvest Publishers, Minsk, 2004; F.G. Kolomoyets, Osnovy sistemnogo analiza i teorii prinyatiya resheniy: Posobiye dlya nauchnykh rabotnikov, uchashchikhsya vuzov i praktikov, Tesei Publishers, Minsk, 2005.

3. Filosofskiy entsiklopedicheskiy slovar, Moscow, 1989, p. 587.

4. T. Saati, K. Kerns, Analiticheskoye planirovaniye, Radio i svyaz Publishers, Moscow, 1991.

5. H. Taha, Vvedeniye v issledovaniye operatsiy, in two books, Book 1, Transl. from the English, Mir Publishers, Moscow, 1985.

6. E.V. Shikin, A.G. Chkhartishvili, Matematicheskiye metody i modeli v upravleniyi: Uch. posobiye, Delo Publishers, Moscow, 2002.

7. H. Raiffa, Analiz resheniy (vvedeniye v problemu vybora v usloviyakh neopredelyonnosti), Nauka Publishers, Moscow, 1997, p. 387.

Col. F.G. KOLOMOYETS (Ret.)

Candidate of Military Sciences