Research Of Web Text Named Entity Recognition Based On Conditional Random Fields

With the rapid development of the Internet, the way of communication is more and more tend to the network. Example micro-blog, network news, comment, etc. They are all belong to the Internet text and the form diversification of the web text. How to get the effective information from the text is a question.That's important.The named entity recognition is a kind of information extraction.Named entity recognition is a key technology of Natural Language Processing, which is a key technology in the text, and it is also an important foundation of text information processing. Currently, there are three methods for named entity. Based on the statistical method and based on rules and based on statistical and rules. The three methods are different, how to make good use of the three methods for named entity recognition is the key.A method for the named entity recognition based on the combining of statistic and rules is put forward. First of the text are standardized, then use CRF model with words and part of speech as features for training. Combined with the daily, random and entertainment of web text and a number of matching rules to named entity recognition. The experimental results show that the method can effectively improve the named entity recognition.