Analysis and Synthesis of Technology for Textual Information Classification

The task of developing effective text information classification systems requires the thoughtful analysis and synthesis variable components technology. These strongly affect practical efficiency requirements to data. For this purpose, a typical technology was discussed, comparing regular “learning from features” approach versus more advanced “deep learning” approach, that studies In order implement technology, first tested, which included means (methods, algorithms) for features source text, by applying dimensionality transformation, building model solutions allow correct data set features. As result, all steps are described, allowed determine way presenting in terms hidden data, their presentation standard visual form evaluate solution, as well its efficiency, based on depth study, informational core document studied, using regression T-stochastic grouping reduction.The separate results contain estimation algorithms time relative performance each step proposed This gives possibility obtain best algorithm intelligent processing is useful given dataset application. estimate suited separation reduced dimension an experiment carried out selection range algorithms, particular boosting methods. result necessary were discussed real conducted, identify most important stages classification.