Incident Classification Program

This paper is dedicated to program for classifying incidents in the Service Desk. The program allows automate the process of classifying incident on all levels of support. Introduction of such program in the processes of Service Desk gives next advantages: reduces the burden on people; reduces the cost of support; makes process of handling of incident faster; improves the quality of service and, as a result, leads to an increase in user satisfaction from the provided services; allows you to fully automate handling of some routine requests. The main purpose of the work is to develop an incident classification program that has an accuracy of classification not less than that of a person. The main tasks need to be addressed in the course of the work: to study the subject area and existing solutions; to study and choose algorithms for classification; to implement and test the program. In the paper machine learning algorithms solving the problem of text classification are described. Text processing methods, term weighting methods and text classification algorithm are also reviewed. In addition, various metrics for evaluating of classifiers are reviewed. They are necessary in order to compare different classifiers among themselves. Based on the obtained results, a method for classifying incidents using existing algorithm is proposed. The program is developed and implemented based on the proposed method.