Abstract: In order to extract musical entities from different Web pages quickly and correctly, this paper presented a hybrid approach based on rules and statistics for Chinese named entity recognition in music domain based on the characteristics in music domain, and implemented the musical named entity recognition system. The experimental results show that this system has a higher precision and recall rate.

Key words: Named Entity Recognition, Musical Named Entity, Hidden Markov Model

摘要： 为了能快速、准确地将分散在Web网页中的音乐实体抽取出来,在全方位了解音乐领域中命名实体的特征的基础上,提出了一种规则与统计相结合的中文音乐实体识别方法,并实现了音乐命名实体识别系统。通过测试发现,该系统具有较高的准确率和召回率。

关键词: 命名实体识别, 音乐命名实体, 隐马尔科夫模型