关键词: 信息抽取, 自由文本, 隐马尔可夫模型, 依存分析

Abstract: Information extraction is an important part of text information processing. The current information extraction researches mostly focus on semi-structured text. It proposes a novel text information extraction algorithm based on the combination of dependency parsing and HMM. The algorithm formulates appropriate rules based on applying dependency parsing to shallow syntactic analysis of sentences, forming the input sequence of HMM to achieve free text information extraction combining the advantage of easily building, good adaptability and high extraction accuracy of HMM. Experimental results show that the new algorithm has very good performance on recall rate, accuracy and correct rate.

Key words: information extraction, free text, Hidden Markov Model（HMM）, dependency parsing