关键词: 信息抽取, 内在认知机理, 命名实体识别, 共指消解, 机器学习

Abstract: Under the guidance of the Knowledge Discovery Theory based on Inner Cognitive Mechanism（KDTICM），this paper focuses on the difficult points of the Chinese named entity recognition.It takes into full consideration the role that expert knowledge plays in the named entity recognition.According to the different types of entities，this paper flexibly combines the method based on statistics and rules.A series of techniques are adopted to deal with the tasks of information extraction，such as machine learning，document understanding and analysis，parsing technique，graph algorithm and graph mining，computing with words，rapid full-text retrieval，etc.The object of exploration of this paper is to obtain from text not only the relations in simple sentences but also the entity relations across sentences and paragraphs.

Key words: information extraction, inner cognitive mechanism, name entity recognition, anaphora resolution, machine learning