A Hybrid scheme for Parsing Cantonese Text Based on PyCantonese Plus and PyLTP

A novel scheme for the Cantonese text parsing based on the method of dynamically expanding corpus is presented. Scheme need prepare an additional Chat format file, i.e., export a Cantonese phase file from Cantonese Dictionary firstly, which is widely used in Chinese Mainland, then annotate Jyutping romanization online and annotate them with part-of-speech (commonly known as POS) tags, finally generate the additional Chat format file based on CHILDES database and the above-mentioned Cantonese phase file Jyutping file, POS file, etc. When parsing Cantonese text, we execute iterative operations such as word segmentation with PyCantonese, expanding corpus until the segment result is ideal. Then we do other tasks including POS tagging with our POS dictionary and PyCantonese‘s pos-tagging module, dependency parsing with PyLTP, and visual display all the above parsing results, etc. Test results show the superior performance of the scheme and potential for the parsing Cantonese text.