Towards Programming Languages for Machine Learning and Data Mining (Extended Abstract)

Today there is only little support for developing software that incorporates a machine learning or a data mining component. To alleviate this situation, we propose to develop programming languages for machine learning and data mining. We also argue that such languages should be declarative and should be based on constraint programming modeling principles. In this way, one could declaratively specify the problem of machine learning or data mining problem of interest in a high-level modeling language and then translate it into a constraint satisfaction or optimization problem, which could then be solved using particular solvers. These ideas are illustrated on problems of constraint-based item-set and pattern set mining.