A syntactic analysis is core technology of the application area using natural language such as Information Retrieval, Question and Answer, Machine Translation and all that sort of things. However, syntactic analysis is not presently usable to practica...
A syntactic analysis is core technology of the application area using natural language such as Information Retrieval, Question and Answer, Machine Translation and all that sort of things. However, syntactic analysis is not presently usable to practical use due to unsatisfactory speed and accuracy. Thus, this thesis proposes a method of improving them.
In an effort to increase speed in syntactic analysis, this thesis presents new parsing algorithm. This is called HCS(Head Candidate Set) parsing algorithm due to utilizing sets of head candidate of each word in parsing process. HCS parsing algorithm is 3 to 11 times faster than CYK parsing algorithm, which is generally used in syntactic analysis.
In addition, there have been two technique invoked for improving accuracy of Syntactic analysis. Applying chunking result is one of them, the other is that takes advantage of morphemes’ statistics rules. Chunking rules are extracted from dependency structured corpus by bi-gram to five-gram of words’ part of speech tag. This method which shows phrases’ inner part of dependency relationship differs with existing study. Morphemes’ statistics rules are statistical information by morphemes that appears in dependency structured corpus. To enhance accuracy, grammatical features of morphemes’ statistics rules are reflected.
，韩语毕业论文，韩语论文网站 |
