Automatic Extraction Method for Specific Domain Terms Based on Structural Features and Mutual Information

In this paper, the abstracts and keywords in the field of aviation literatures are used as corpus. Firstly, analyze the word formation characteristics of domain term in corpus, these characteristics are used to construct the part-of-speech combination model that adapted to the specific domain. And then, the distance between two adjacent word tokens in the candidate terms was calculated with mutual information, and set the threshold to convert the term extraction into a classification problem of whether the candidate term is domain specific. Finally, experiment result show that the method can be effectively applied in the automatic extraction of the term.