Атты І халықаралық конференция ЕҢбектері
жүктеу/скачать 8,57 Mb. Pdf көрінісі
|
|
2 Related works There was a variety of techniques used for phrase recognition. Which include rule-based technique, statistical technique, or a combination of rule-based and statistical techniques. For example, Church's English Base NP Recognition(1988) [1]. His approach is through manual or semi-automatic annotation phrase corpus as a training corpus, then any pair of speech tags phrase context information in the statistics of probability, which according to the above probability words in the sentence made between any two adjacent markers, to obtain analysis results. This approach followed in shallow parsing has been widely used. Moreover, several main approaches or algorithms to phrase recognition was investigated, typically implemented using a Chunk parsing for statistics model to decide the boundary[3] . Chunk 49 parsing were first introduced by Abney (1991)[2], which is the most widely used syntactic parsing. The main idea of chunk parsing lies in seeking the appropriate breakthrough point, and decomposing the full parsing problems to syntax topology statistics structure and syntactic relation. Jun Zhao and Changning Huang are pioneers in Chinese phrase studies (1998)[4]; Tsinghua University had also completed its TCT (Tsinghua Chinese Trebank ) for Chinese (Qiang Zhou,2004)[5]; many language studies of it had been used Kazakh phrase recognition parsing(Gulila.A etc.,2009))[6-7] . Maximum Entropy was first introduced to NLP area by Berger, et al (1996) and Della Pietra, et al. 1997[8].which is an extremely flexible technique for linguistic modelling , since it can use a virtually unrestricted and rich feature set in the framework of a probability model. It is a conditional, discriminative model and allows for mutually dependent variables[9]. жүктеу/скачать 8,57 Mb. Достарыңызбен бөлісу:
|