基于小句相似度计算的专有名词识别

Proper Nouns Recognition Based on Sentence Similarity Computing

专有名词识别是中文信息处理领域的一个难题。句子相似度计算方法在中文信息处理领域有着广泛的应用。本文探索性地使用句子相似度计算方法来解决专有名词识别问题，并针对专有名词识别的研究背景对小句相似度计算方法做了一些改进，改进的计算方法不仅考虑了公共字符，而且还加入了语义信息和结构信息。实验证明该方法是可行的。

Proper Nouns Recognition （PNR） is a difficult problem in Chinese information processing. Sentence similarity computing has been widely used in Chinese information processing nowadays. This paper introduces a method of PNR based on sentence similarity computing, and proposes an improved sentence similarity computing method according to studying PNR. Combining the characteristics of corpus, this method not only considers the common string of the sentences, but also uses semantic information （using HOWNET） and structure feathers. The experimental results show that this method is satisfying.