Automated categorization of video lectures

Jernej Virag (2012) Automated categorization of video lectures. EngD thesis.

Preview

PDF Download (1603Kb)

Abstract

This bachelor thesis presents service implementation for categorization of video lectures using machine learning methods. The purpose of the service is to automatically categorize lectures on Viidea platform powering VideoLectures.net web portal. This work describes realization of a web service written in C# programming language using Latino machine learning library. The service implements four types of classification algorithms: naive Bayes classifier, k-nearest neighbour classifier, maximum entropy classifier and hierarchical classifier. All classifiers are described in this work. Service implementation consists of four components: data model with lecture data, categorization module, lecture recommendations module and web service interface. The service interface uses HTTP protocol and the service itself runs in .NET CLR and Mono environments. The quality of classifiers is evaluated at the end of this work using two distinct methods.

Actions (login required)