Novelty Detection for Topic Tracking CEM AKSOY 1, FAZLI CAN

Multi-source web news portals provide various advantages such as richness in news content and an opportunity to follow developments from different perspectives. However, in such environments, news variety and quantity can have an overwhelming effect. New event detection and topic tracking studies address this problem. They examine news streams and organize stories according to their events; however, several tracking stories of an event/topic may contain no new information, i.e. no novelty. We study the novelty detection (ND) problem on the tracking news of a particular topic. For this purpose, we build a Turkish ND test collection called BilNov-2005 and propose the usage of three ND methods: a cosine similarity-based method, a language model-based method, and a cover coefficient-based method. For the language model-based ND method, we show that a simpler smoothing approach, Dirichlet smoothing, can have similar performance with a more complex smoothing approach, Shrinkage smoothing. We introduce a baseline that shows the performance of a system with random novelty decisions. Additionally, a category-based threshold learning method is used for the first time in ND literature. The experimental results show that the language model-based ND method significantly outperforms the similarity- and cover coefficient-based methods and category-based threshold learning achieves promising results when compared to general threshold learning