Event element recognition based on improved k-means algorithm

Abstract

Event element recognition is a difficult problem in the extraction of event information, and its important content includes the identification of time and location elements. At present, the identification of time and location elements mainly uses the method of machine learning, but the method based on machine learning is susceptible to sparseness of corpus. An event element recognition method based on improved K-means algorithm is proposed. Firstly, the method is preprocessed to get the annotation corpus, and then the K-means algorithm is improved. The K value of the cluster is obtained by using the Canopy algorithm, and then the clustering analysis is carried out according to the improved algorithm. Finally, Feature filtering unrelated data, and re-use the clustering algorithm to cluster the data, to achieve the event time and location elements of the identification. The experimental results show that the method has achieved good results, and the F value of time and location recognition is 73.82% and 72.38%.