Construction of Information-Analytical Platform for Scenario Analysis Based on Large Volumes of Weakly Structured Data
The research was aimed at creating an information-analytical platform that provides the conduct of scenario analysis for social, economic, ecological processes in complex systems with human factor, using deep analysis methods of large volumes of weakly structured data, aggregated from different sources, and expert approaches of
foresight methodology. The developed information-analytical platform provides collection, aggregating, processing, classification of heterogeneous structured and unstructured information on a given problem, highlighting the main indicators and predicting their behavior, employing qualitative analysis methods to cope with incompleteness, lack of precision, fuzziness, contradictive nature of present data, determining critical factors, foresight of social, economic, ecological processes in complex systems and constructing scenarios regarding them.
The modern approaches, methods and models for storing large volumes of weakly structured data were studied. The potential sources of weakly structured data, as well as types of documents from these sources, are studied and classified. New techniques for solving the problem of extracting ratings, opinions and expectations, as well as emotional coloring of situations described with natural language are introduced, providing decision-making support. A multi-component information model structure for extraction and analysis of text emotional coloring was implemented and tested. A method for assessing and improving quality of expert assessment in
analytical hierarchy method was developed. A technique for modeling realistic expert opinions, where the uncertainty is stipulated only by Saaty scale, pessimistic and optimistic expert opinions, that have additional uncertainty related to these qualities, was suggested. The criteria for quality of input data describing their uncertainty for modified morphological analysis method, were developed. A number of new techniques and procedures for modified morphological analysis method were suggested, improving the method results when using weakly structured information as input data/ A fuzzy modified morphological analysis method was created. All the developed within the research methods, procedures and algorithms were implemented in a complex software tool set.