Classifying the Hungarian Web

In this paper we present some lessons learned from building vizsla, the keyword search and topic classification system used on the largest Hungarian portal, [origo.hu]. Based on a simple statistical language model, and the largescale supporting evidence from vizsla, we argue that in topic classification only positive evidence matters.

• Type: Conference
• Year: 2003
• Where: EACL
• Authors: András Kornai, Marc Krellenstein, Michael Mulligan, David Twomey, Fruzsina Veress, Alec Wysoker
• Keywords: EACL 2003, Largescale Supporting Evidence, Natural Language Processing, Simple Statistical Language, Topic Classification