A Bootstrapping Approach for Chinese Main Verb Identification

The task of main verb identification is to recognize the predicate-verb in a sentence. This task plays a crucial role in various areas such as knowledge acquisition, text mining, and question answering, and is also an important preprocessing for many applications including sentence pattern analysis and semantic roles identification. This paper proposes a domain-independent bootstrapping method to automatically identify main verbs of sentences from unannotated domain-specific Chinese unstructured texts. Experimental results in two domains show that the algorithm is promising. As applications of the main verb identification, we have developed a main verb driven approach of extracting domain-specific terms from unstructured text corpus.