FAQ

Q:For every system function, is the input file always the original Chinese file?
A:Not always. All the functions NiuParser provides are interdependent. That is, for some function, the input files have to be the processed results. Specifically speaking,
      The input files of the word segmentation are the original Chinese data files.
      The input files of the POS tagging are the output files which are processed by word segmentation.
      The input files of the named entity recognition are the output files which are processed by POS tagging.
      The chunking input files are the output files which are processed by POS tagging.
      The Constituent parsing input files are the output files which are processed by POS tagging.
      The Dependency parsing input files are the output files which are processed by POS tagging.
      The Semantic role labeing input files are the output files which are processed by constituent parsing.
Q:What encoded files does NiuParser support?
A:Currently, NiuParser only supports the files encoded with UTF-8. If the encoded file is not UTF-8, the system will display the error message telling illegal encoding.
Q:Do the original Chinese files have to be the format of one sentence per line?
A:Yes. The present version of NiuParser only supports the file format of one sentence per line. That is, NiuParser doesn't support textual level processing or it does not provide sentence segmentation function. If the original data is a passage, it should be divided into sentences. Only converting the text into the format of one sentence per line can it be processed by NiuParser.
Q:Does the word segmentation function of NiuParser support user dictionary?
A:Yes. For specific usage, please refer to NiuParser Manual.
Q:I have my own word segmentation and POS tagging system, and I just want to use dependency parsing function of NiuParser. Does it work?
A:Yes. All the functions NiuParser provides are relatively independent.
      If users want to use dependency parsing function of NiuParser, they need to convert the format of the POS tagging results into the POS tagging format required by NiuParser.
      Note that the input POS tagging results must meet the standard of the Penn Chinese Treebank.
      For other functions, the principle is similar. That is, users needs to convert the their input into the format required by NiuParser. For specific format specifications, please refer to NiuParser Manual.
Q:After completion of constituent parsing by NiuParser, when conducting semantic role labeling for its results, why does the system display "Constituent Parser tree format error in line ***"?
A:The reason for the error is that the format of the constituent syntax tree input is illegal. The possible reason is failure of constituent parsing, that is, it outputs a syntax forest instead of a syntax tree. It may be solved by adding brackets at the outermost layer.