A processing method of class-2 unknown Japanese compound words of two components with use of in-word grammar and its prototype system implementation

Techniques of natural language processing are applied to many kinds of systems. But no systems have the ability to process unknown words. Especially in Japanese, it is essential for systems to have such abilities. A method using in-word context is proposed to process class-2 unknown Japanese compound words of two components. First, unknown Japanese words are classified into three classes. Then, in-word grammar for compound words of two components is organized to build a lexicon. A system is implemented to evaluate the proposed method