The thesis research object is the decision-making support (DS) process for weakly-structured complex systems. The purpose of research is to increase the reliability of recommendations provided to decision makers (DM'), through use of DS tools based on theoretical and technological principles of expert DS, developed by the author and allowing to make domain models more consistent with experts' own views. The thesis uses a system methodology, methods and models of information theory, artificial intelligence, methods of the theory of matrices, graphs, sets, mathematical statistics, game theory, etc. In the thesis a new model for obtaining expert information is suggested for the first time. Based on this model, an expert estimation technology using scales of different informativeness level is developed. It allows to avoid the distortion of information obtained from experts. After gradual interactive increase of scale informativeness level, the final estimation is carried out in the scale, which is most consistent with the actual degree of expert's awareness of the subject matter. An aggregation method is suggested for group incomplete pair-wise comparisons defined in different scales. Due to its single-phase character, the method allows to apply feedback with experts in order to achieve sufficient consistency - feedback algorithm is also developed for the first time. New methods for defining and achieving sufficient ordinal expert estimation consistency level, which allow to fully utilize these estimates in the group examination process, are developed. A set of original group cardinal expert estimate processing methods is proposed. In these methods, when pair-wise comparisons are aggregated, the information derived from all spanning trees of the graph corresponding to the pair-wise comparisons matrix is taken into account. Based on the proposed new models and methods, a DS technology, which allows to evaluate decision variants according to expert-built subject domain model, is designed. DS software products "Consensus" and "Solon-3", where the proposed technology is implemented, have found practical application during group expert examinations in the Situational Centre of the Institute of Mathematical Machines and Systems of the NASU. Personnel evaluation system "Supervision" is used in the Russian Federation supervisory body "Rostehnadzor" to assess inspectors and structural subdivisions, conducting environmental, technological and atomic supervision of the functioning of potentially dangerous objects, and implemented by "MKKU-merezhi" Ltd. in the government of Ukraine.
