Semantic networks

Last years the great attention began to be given to research of problems of management by various social systems, including – higher education. This objective process is caused by dominating influence of some circumstances. First, under the influence of radical reforms not so much direction has essentially changed, leading factors of social processes in a modern society how many have changed. For example, if earlier the country population acted as object of administrative influence subjective activity of the citizens choosing this or that form of formation or a direction of training now has considerably grown. Secondly, destruction of former structures of management by higher education has raised the question not only about perfection former, but also about working out of essentially new control systems, with the account of constantly changing, self-adjustable character of a development of education.

Main objectives and values of higher education of The Republic of Belarus are set by the Constitution of The Republic of Belarus, laws and other legal certificates. However they are concretized with a view of and values of subjects of educational process. The given purposes and values always combine objective and subjective. They do not operate in itself, will not be understood yet by people, accepted by them internally, translated into language of practice and embodied in activity. Social values act and in the form of a certain complex of requirements to activity, but specific features of people and feature of conditions serve as the reason of occurrence selective, a prejudice to them: something is approved and accepted, and something is not present, something admits to more important, than should be, and something on the contrary, something is thought out (is not provided absolutely not standard) etc.

Macroindicators (the public purposes and the interests expressed in the state ideology in a sphere of education, government programs, standard documents), mesoindicators (internal parameters of the higher school as institute – all establishments and the organisations providing educational process) and microindicators (value and purposeful structures – valuable orientations and the purposes of subjects of educational process), making a basis of the social and management mechanism in a control system of higher education, define character and an orientation of social activity of all subjects of educational process.

Value and purposeful structures of activity can be studied by various sociological means. However in administrative practice it is necessary to consider that the information in the mass-media, containing various estimations of accepted administrative decisions, can serve as quite adequate compound component entering into structure of the social and management mechanism. Techniques of the content-analysis of materials of mass-media allow to carry out and necessary quantification of the substantial parameters. The analysis of mass-media is an accessible and widespread procedure which can act as an element of the sociological examination accessible at any administrative level.

The analysis of the public opinion presented to mass-media, can do not become simple an element, and a basis of sociological examination of accepted administrative decisions in which result come to light social and cultural features and social expectations of all subjects included in system of higher education. Such examination as a matter of fact to be spent mass-media, but its results not always and are not to the full used with a view of effective realization of a state policy in a sphere of education.

Now the particular interest for the social and management mechanism of formation of higher education of The Republic of Belarus, besides the press content-analysis, can represent and technology of the analysis and visualization of the maintenance of the text on the basis of the semantic networks, intended for the decision of the expert problems connected with search of the latent information in the big document files.

At the heart of the given technology, used by us for the analysis of documents, representation of sense of the text in the form of an associative semantic network lies. A semantic network – the information model of the subject domain which are looking like the focused count, which top correspond to objects of a subject domain, and arches (edge) set relations between them. Concepts, events, properties, processes can be objects. Thus, the semantic network is one of ways of representation of knowledge. In the name terms from two sciences are connected: semantics in linguistics studies sense of units of language, and the network in the mathematician represents a version of the count – a set of the tops connected by arches (edges). In a semantic network the role of tops is carried out by concepts of the knowledge base, and arches set relations between them. Thus, the semantic network reflects semantics of a subject domain in the form of concepts and relations.

The semantic network joins most often meeting words of the text which bear the basic semantic loading. For each concept the set of associative communications is formed, i.e. the list of other concepts, of a combination with which it met in text offers.

Original linguistic algorithms use morphological and parse, and also the thesaurus of Russian for an identification of faithful words and word-combinations. For example, such expressions as «the subject of management in higher education system» and «an operating subsystem in the higher school», are led to one concept «the subject of management in higher education». Besides, from among concepts are excluded common words which do not bear independent semantic loading or have wide value. So, words "concept" and "reforming" in itself are not the concepts characterizing the maintenance of the document, but can form the concept expressed by a combination: «the concept of reforming of higher education».

Each concept is offered to be considered as a name of a corresponding theme of the document. The statistical data about communications of concepts of the text and syntactic roles of concepts of phrases allows to estimate their contribution to the general maintenance of the document and to range themes on information. As a result of semantic network everyone to a theme the so-called thematic weight is appropriated.

The maximum value of thematic weight (equal 100) corresponds to a key (major) theme of the document. Value of weight of a theme close to zero shows that it only is casually mentioned in the text, and in it there are not enough the data concerning the given theme.

The semantic network represents a thematic index of analyzed texts which is used for search of documents in themes and their communications, and also for inquiry expansion is associative the connected themes. On each of network themes the set of offers of the text - the citations concerning corresponding theme which represent the thematic resume (abstract) of the text is formed. Besides, ranging of these offers on scales (from 0 to 100) which reflect their information for a corresponding theme is carried out and allow to choose in the resume only the most informative offers. The general resume of the text is formed of the most informative fragments on key themes of the document. Knowledge of associative communications allows to reveal the latent dependences between the objects-themes, interesting events, a personnel, the organizations, etc.

Work with a semantic network can help the sociologist with the decision of following problems: 1) research of thematic structure of a target collection of documents - for example, a news stream for the chosen interval of time. Revealing of key themes and their coherent congestions (semantic fields), being in focus of attention and influencing situation development. Monitoring of dynamics of an information stream in time in terms of capacity of key themes and their communications; 2) search of the new, unexpected information connected with investigated object-theme. The communications hidden in documents with other objects (persons, the organizations, events), revealed in a semantic network, and also the associative chains connecting set objects can represent interest; 3) revealing in documents of reinforcements of known and unknown communications between objects-themes. Search of the concrete documents opening interesting communications, allows to receive detailed representation about character of relations between objects.

For the analysis and visualization of the maintenance of the text on the basis of semantic networks program Russian Context Optimizer Fact Extractor which can be applied sociologists at the organization of administrative process in system of higher education of The Republic of Belarus has been created.

Russian Context Optimizer Fact Extractor – the program allocated with intellectual possibilities intended for the analysis of the text in Russian. She allows to find descriptions of the facts of the set type and to make their classification and ordering. The basic scope of the program – the analytical problems arising at carrying out of computer investigation and demanding high-precision information search, including automatic selection of a material to the file on target object or monitoring of the certain parties of its activity shined in mass-media. The most advanced technologies of an artificial intellect developed by company "Guarantor-park-Internet" are embodied in program Fact Extractor, known for the decisions in the field of computer linguistics.

However semantic networks can render considerable aid not only at the analysis and information search, but also at direct management of higher education in The Republic of Belarus.

At use of semantic networks in administrative process in higher education system the operating system breaks into three levels: supervision, identifications and managements. On each of these levels the dynamic semantic network (DSN) is used.

The dynamic semantic network offered by Smolentsev S.V., (Dynamical Semantic Network - DSN) – a network which not only allows to store knowledge, but also itself processes them, solving all range of problems, characteristic for real intellectual systems, thereby to a certain extent models their functions.

At supervision level the problem of construction of the description of a condition of conditional object of management on signals from sensor controls (all subjects of educational process) dares. For its decision the observer - special DSN , contacting to sensor controls through a touch field of the knots is used.

At identification level the carrying out problem under the received description of identification of a current condition of conditional object of management dares. The problem dares with use of the identifier realized as hierarchical DSN .

At management level problems of an estimation of a current condition of conditional object of management and definition of management for object reduction in the set condition are solved. At this level it is realized DSN , named a regulator. For construction of the latent dependence of estimations of quality from an object condition the algorithm of processing of the information is used. The problem of search of management is reduced to a problem of construction of sequence of the actions translating conditional object of management from a current condition through intermediate conditions in the target.

On the basis of use of dynamic semantic networks probably to build the intellectual control systems capable not only to solve management problems, but also to self-learn. As the operating system is broken into three levels updatings and perfection can be made independently at each level. Control systems on the basis of DSN can be trained and arranged under features of various objects of management in various subject domains.