Identifying Topics Based on Event Sequence Using Co-occurrence Words
Kei Wakabayashi, Takao Miura · 2008
In this paper, we propose a sophisticated technique for topic identification of documents based on event sequences using co-occurrence words. There have been many investigations for document classification based on vector space modeling, but here we consider each document as an event sequence each event as a verb and words correlated with the verb. We propose a new method for topic classification of event sequences by using Markov stochastic process modeling. We show some experimenal results to examine the method.