|
Syntactic analysis is a very important research in natural language processing . The analysis of a language , including lexical analysis, parsing , semantic and pragmatic analysis of several levels , parsing a direct impact on the process of subsequent works . The thesis established treebank annotation system , the Uighur statement , this labeling system includes two modules: the function chunk Mark modules and components marked module . By this labeling system , we completed the basic work on the establishment of the rule base , we analyze the syntactic analysis method is popular at home and abroad , choose Chart parsing method , beginning with a bottom-up Chart parsing method However, when the efficiency of this method of analysis is relatively low , and prone to ambiguity , we improved the algorithm , the bottom-up and top-down Chart algorithm combined to achieve a rule-based Uighur Chart parser . Conducted on the basis of the analyzer test multilingual IT Autonomous Key Laboratory of parts of speech in the Xinjiang University Corpus (XJU UPOS Corpus) , test results are better . Chart System Analysis , we found some problems, such as: the smaller of the selected dictionary and rule base , between the rules in the rule base often conflict . Because we know that natural language there is ambiguity , resulting in the above-mentioned problem is inevitable . Future work should continue to improve the syntactic treebank annotation system , thus improving the accuracy of the rule base , and the syntax rules of the rule base should be more complicated and complete , thereby improving the efficiency of the Chart parsing decency .
|