关键词: 关键词抽取, TFR*IDF, 多特征, BP神经网络

Abstract: Chinese keyword extraction is a worthy problem for study in natural language understanding. By using the TFR*IDF, part of speech, the first of the current position of features as the extraction characteristics of the Chinese document keywords extraction, while features of weight trained by the BP neural network, eventually it can get Chinese keyword. By analog simulation, it is proved that the method is effective.

Key words: keyword extraction, TFR*IDF, multi-features, BP neural network