Building Chinese Semantic Treebank For Patent Text On The Basis Of 3 Dimensional Dynamic Concept Model

In this paper we proposed a method for building Chinese semantic treebank for patent text. On the basis of 3 Dimensional Dynamic Concept Model, we designed the semantic relations tagging sets with chunk as tagging object for patent translation. We proposed a top-down process for manually tagging and a standard for processing treebank, which is used for the level partition of a sentence and also can reflect core words. Building semantic treebank for patent text is very important for fully exploiting useful information of the patent text.