Chinese word sense disambiguation based on extension theory

This paper applies extension theory to make extension transform like extension decompose and extension replacement for ambiguous words. Having established relevant extension set for every sense of ambiguous words, we acquire automatically the linguistic knowledge about ambiguous words from the large scale corpus by the sense words in the extension sets, and build the unsupervised word sense disambiguation model. The method overcomes the difficulty to learn WSD knowledge from the untagging corpus. Experimental results prove that our technique is available for Chinese WSD and the average accuracy reaches 90.16%.