基于神经网络的中文姓名抽取技术

设计了一个中文姓名抽取系统, 该系统采用神经网络进行汉语句子的分词处理, 根据姓名后置特征词进行姓名的抽取, 成功解决了尾字和下文成词的姓名抽取问题. 以1998年1月份《人民日报》语料库中含有此类姓名的语句作为测试数据,结果表明, 姓名抽取的召回率和精确度较现有方法都有很大提高.

Extraction Technology of Chinese Names Based on Neural Network

An extraction system of Chinese names was designed. The system adopts the neural network to deal with the Chinese word segmentation, then carries on extraction of name by means of rearmounted characteristic word according to name. The system solves the question of extracting Chinese name succes sfully whose tail character and the following character construct a word. With the sentences with this kind of name in “People’s Daily”’s corpus base of Janua ry of 1998 the testing data, the result shows that the recall and precision rates are all improved a lot compared with existing methods.