Abstract

The paper discusses the problems in the detection of security information in legal and other administrative data. The authors analyse the prospects of the use of datamining in the solution of two key problems: abundance and indirectness of these data. Security research uses two kinds of data. The first one is scientific data, designed and gathered specially for the verification of certain security theories. They are the data of criminological, sociological, psychological surveys, experimental data, etc. The second kind is data that are not designed for security research. Most of the data are the by-products of the national legal system, especially its criminal justice, of national and local institutions responsible for the maintenance of public order, and of other public agencies. All the data provide huge amounts of information destined to control and direct the activities of these institutions. There are two kinds of problems in the use of this information in security study. First, this information has to be ‘decoded’ from data describing the activities of related institutions. Second, data on security are lost in huge amounts of other public data. Thus, they have to be ‘mined’. The paper discusses the prospects of the modern information proceeding method—data-mining in the solution of both problems. A new concept of data mining as a meta-procedure is proposed. The object of this meta-procedure is supposed to be the integration of multiple current theories (for example, criminal security) and related statistical procedures. A general algorithm of such data-mining is proposed