A Maximum Entropy Model Application on Recognition of Metaphor Phenomena

Metaphor happening in our daily life language is a usual phenomenon, and recognizing them by the use of computer is becoming a valuable research task in the field of natural language processing, artificial intelligence, and even applied linguistics. This paper proposes a way to recognize metaphors based on maximum entropy model after analyzing the features of metaphor, and reasons the rationality to build a recognition model using the statistical methods. The results of the experiment show that the model performs well at high precision and recall, as well as the f value, thus we come to the conclusion that such metaphor recognizing model based on maximum entropy principle has a promising future