Chunk Based Chinese Syntactic Parsing And Its Application

As one of the most important technologies of natural language processing,syntactic parsing is the foundation of many natural language processing tasks.It is particularly important for language processing tasks that require the use of deep sentence structure.Therefore,syntactic parsing technology has been widely studied.However,there still need to be more promotion for this technologies than others due to its difficulty.We should do further work on it,let alone the uniqueness of its much more free syntactic structure.In this paper,we carry out some work based on the characteristics of several former work of syntactic parsing.We design a chunk based Chinese syntactic parsing in this paper.We divide into three stages.First,sentences are divided into chunks.Then we get the chunks parsed.Finally we merge all the parsed chunks into a whole syntax tree.During chunking stage,a conditional random field model is implemented by taking chunking as a sequence labeling task.In the chunks syntactic parsing stage,it is divided into two sub module which consists of chunks structure analysis and label classification in order to reduce the complexity.We implemented a chunks structure analysis algorithm based on conditional random field model by transforming it into a sequence labeling problem.And beam search is used to alleviate the error accumulation problem when parsing.We make a classifier for syntactic tag classification based on conditional random field model by using rich features.The experimental results show the effectiveness of this Chinese syntactic parsing addressed by this paper.Finally,we apply this syntactic parsing to an information retrieval system based on tourism.It shows that and this syntactic parsing just bring a lot of improvement to the system.