Interpretable models constructed by machine learning algorithms from complex domains often contain relations that are statistically significant, but meaningless to a human. We propose a novel method, named Human-Machine Data Mining (HMDM), which combines human understanding and computer data mining methods to extract relations that are meaningful to the human and statistically supported with data. We term such relations credible relations. The HMDM method defines a procedure and a toolbox that human uses in interactive and iterative manner to direct computer search towards those parts of the search space with credible relations. We applied HMDM to macroeconomic, demographic and web genre classification domains.
Searching for credible relations through interactive data mining – Information Sciences, 2014
HMDM software tool is a Java desktop application.
The method is applied to macroeconomic data to find credible relations between the level of high-level knowledge sector development and economic welfare of a country. The results of analysis are presented in the paper and data is available for download.
User study was designed to evaluate HMDM. The questionnaire, results and data used to build the models from questionnaire are available for download.