Methodology: Discrete mathematics, computational logic, formal concept analysis, machine learning, data mining, computer linguistics, ontology modeling, theory of algorithms, software engineering.
Empirical base of research: Data sets in the form of relational databases, collections of texts in natural languages, structural data in the form of graph models of different types.
Results of research:
1. The analysis of the bibliography and completion of reviews on areas of formal concept analysis, recommender systems, ontological modeling, and accessibility of freely available data sources. Within the theoretical research in FCA, clustering, and text processing the laboratory team has accumulated a significant amount of benchmarking datasets.
2. The increased efficiency of the implementation of basic FCA algorithms for construction formal concept lattices and calculation stability indices of formal concepts; the implementation has been used in the tasks of medical informatics. Development of new versions of methods and algorithms for clustering and classification on tricontexts; the implementation has been tested and used in online recommender systems.
3. Development of the prototype of the original software component proposed in 2013 to work with pattern structures. It is integrated with the means of solving the problems of classification, which allowed approaching the creation of a universal system for studying problems of classification on the basis of diverse complex attributes, including interval structures, sequences and graphs.
4. Recent modifications of DOD-DMS platform that simplifies the construction of the scientific and applied software systems in the field of data analysis, especially in pre-processing of data from external sources, local storage of complex data, and efficient indexing of natural language texts. Updates of several subsystems of the automated research system FCART (Formal Concept Analysis Research Toolbox), which aimed at FCA researchers and related areas of discrete mathematics and data analysis, have been performed. A refined set of tools features tools for analyzing indexes formal concepts of any kind, tools for pattern structures processing, report editor, and built-in scripting language.
5. A methodological and technological basis for processing "big data" in the Internet. Several variants of systems to maintain access to heterogeneous data sources have been proposed. The second version of the local data storage subsystem of FCART including a new scheme of user authorization has been created. Using the local storage subsystem, a prototype to work with open data sources has been implemented.
Level of implementation, recommendations on implementation or outcomes of the implementation of the results: the application of methods for development of intelligent systems; the use of the developed software to analyze complex data in various fields (the testing usage has been conducted in the following domains: health care, law enforcement, e-commerce and Internet marketing).