Text categorization with machine learning and hierarchical structures
M. Krendzelak, František Jakab · 2015
Text categorization with machine learning algorithms usually assumes to have flat set of categories. Such classifiers are very domain specific and not reusable for some other generic text classifications. It is very possible that a hierarchically structured set of categories might have a higher impact on the way classifiers are used and built. As presented in this document, the list of most common approaches for text categorization disregards hierarchy. The main reason is because measured performance of hierarchical trials has not showed a significant difference. Nerveless, it is encouraged to perform additional research in text categorization with hierarchical structures.