Event recognition is one of the most fundamental and critical field in information extraction. In this paper, Event recognition task can be divided into two sub-problems containing candidate event triggers identification and the classification of candidate event trigger words. Firstly, we use trigger vocabulary generated by trigger expansion to identify candidate event trigger, and then input sequences are generated according to the following three features: word embedding, POS (part of speech) and DP (dependency parsing). Finally multiclass classifier based on joint neural networks is introduced in the step of candidate trigger classification. The experiments in CEC (Chinese Emergency Corpus) have shown the superiority of our proposal model with a maximum F-measure of 80.55%.
CITATION STYLE
Liu, W., Yang, Z., & Liu, Z. (2018). Chinese event recognition via ensemble model. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11305 LNCS, pp. 255–264). Springer Verlag. https://doi.org/10.1007/978-3-030-04221-9_23
Mendeley helps you to discover research relevant for your work.