摘要 |
A method for processing a natural language using unification grammar is disclosed. A method for processing a natural language according to an embodiment of the present invention comprises: receiving a natural language text; pre-processing the input natural language text to remove stop words, inflectional affixes, and unregistered words; and connecting features stratified by predefined unification rules with corresponding feature values to generate a stratified feature tree, wherein the feature tree sets the pre-processed natural language text as a root node, generates construction features as child nodes of the root node to assign corresponding construction feature values to the construction features, respectively, and generates morpheme features which are the minimum unit of a language having meaning as child nodes of the construction features to assign corresponding morpheme feature values to the morpheme features, respectively, thereby structuralizing the morpheme features into constructions and construction features into a sentence. [Reference numerals] (210) Input a natural language text; (220) Pre-process the input natural language text; (230) Restore an agglutinative language to its original form before combination; (240) Divide the pre-processed natural language text in a morpheme unit and attach parts of speech; (250) Divide the sentence of the natural language text in a construction unit; (260) Generate a feature tree structuralized by predefined unification rules; (AA) Start; (BB) End |