An Improved Method for Chinese Company Name and Abbreviation Recognition.

Recognizing Chinese company name entity extraction and corresponding abbreviation is vital and challenging in Chinese natural language processing, as traditional methods have difficulties in time sensitive, recognizing Out of Vocabulary words, and collecting training data from open data set etc. To avoid the effects of those issues and improve the recognizing performance, we develop a new algorithm that combines and optimizes the rule-based way, dictionary-way and statistics-learning way together. The precision, recall and F are 96.75%, 89.05% and 92.74% respectively in financial news from the internet.