Automated categorization of video lectures

摘要

This bachelor thesis presents service implementation for categorization of videoudlectures using machine learning methods. The purpose of the service is toudautomatically categorize lectures on Viidea platform powering VideoLectures.net webudportal.udThis work describes realization of a web service written in C# programming languageudusing Latino machine learning library. The service implements four types ofudclassification algorithms: naive Bayes classifier, k-nearest neighbour classifier,udmaximum entropy classifier and hierarchical classifier. All classifiers are described inudthis work.udService implementation consists of four components: data model with lecture data,udcategorization module, lecture recommendations module and web service interface.udThe service interface uses HTTP protocol and the service itself runs in .NET CLR andudMono environments.udThe quality of classifiers is evaluated at the end of this work using two distinctudmethods.ud