Parallel Processing Strategy for Segmentation Ambiguity

In the actual analysis of corpus, it showed that simply using the annotated corpus´s information to detect combinational ambiguity can cause serious understanding bias and make confusion of segmentation inconsistency, while the resolution of real ambiguity is the most difficult point of overlapping ambiguity strings. We propose parallel processing strategy, namely, taking advantage of bilingual dictionaries, translations of parallel corpus and context information to resolve the real overlapping segmentation ambiguity, and setting combinational ambiguity library to distinguish the inconsistency between combinational ambiguity and segmentation.