关键词: 命名实体识别;特征选择;支持向量机, 条件随机域

Abstract:

These methods make extensive use of a diverse set of features, including local features, full text features and the features of external resources according to characteristic of algorithms. All the features are integrated effectively and efficiently into the recognition systems. Also the impact of different feature sets on the performance of the systems is evaluated. In order to improve the performance of systems, a post-processing module is added to deal with the abbreviation phenomena and the cascaded name entities as well as the identification of boundary errors. Evaluations of experimental results prove that the strategies of the feature selection and the post-processing modules have important contributions to better output of the systems.

Key words: name entity recognition, feature selection, support vector machine, conditional random fields