Chinese Information Processing On The Link Writing For Chinese Word

As is known to all, every natural language processing is better based on words, because words are the smallest linguistic units with independent meaning. Chinese texts are only concatenations of characters. Chinese words are not delimited by spaces in texts as that in English. Chinese word segmentation is therefore the first step for any Chinese information processing and hinders seriously its development.Considering the Chinese information procession's development trend, we designed a new Chinese input software with word segmentation, this method can put the phrase of Chinese segmentation into the time when inputting the Chinese words. Based on the theory and practice of major innovations are: analysis the necessity and feasibility of link writing for Chinese word, puts forward the strategy of link writing for Chinese word: introduces Chinese input software for link writing for Chinese word. It proposes a progressive refinement of the segmentation strategy based on the omni-segmentation. First, the strategy use un-cross-border ambiguity algorithm to determine the long Chinese string into a number of sentences, each sentence went to the omni-segmentation algorithm to generate Chinese segmentation directed graph, and then through the phases of certainty segmentation and identification of unknown word, the plan is to reduce the segmentation path. Using graph theory of adjacency matrix as a conversion, a depth-first search to the entire map, can be cut at all possible paths. Finally, the various paths of segmentation comprehensive measure, based on some principles such as: mutual information of the word around the breakpoint, and other information to the terms of each frequency splitting path scoring, selects appropriate segmentation results.