Data Mining is useful tool and source of heuristic information for sociologists today. Nowadays society is much more complex than ever, with strong tendencies to generate new social phenomena, routine practices, patterns of thinking and behavior in all spheres of human activity. At the same time sociology faces a problem of huge increae of data and wider perspectives of information and knowledge extraction from this data (Big Data). All these aspects actualize optimal methodological instruments and sets of skills that permit scientists to analyze current society objectively and effectively. Data Mining requires cross-disciplinary skills to organize this kind of research. In the article we definite Data Mining as a number of mathematical methods of getting new knowledge from large sets of data (sources and volumes) with all special technological specifics for their collecting and processing. There are many Data Mining methods, so we describe most useful for sociological research ones: cluster analysis, decision trees, associate rules, logistic regression and neural networks. Very important part of Data Mining application is Data Warehouse building with operative analysis and visualization using of multy-measures tables (OLAP). Last part of the article covers a case of Data Mining application for analysis of labor market with the aim to describe structure of this object. We demonstrated all steps of data preparation, modeling and interpretation. Results of this analytical project permitted to describe situation in Altai krai labor market with regard to vacancies and applicants. Information gave scientists a unique opportunity to analyze objective side of social institutes, social structure dynamics.