The best application of this repositary is: building a domain specific dictionary using all provided texts in one particular domain.
The amazing thing is that (1) it doesn't require labelled data; (2) it doesn't require an explicit dictionary.
I applied this tool to parse financial tweets from Weibo. The word counts can be used as a reliable financial dictionary for parsing Chinese text using HanLP.
##################### Only requires one input: raw_text_to_be_analyzed.txt (encoding="utf8")
##################### Important note: this tool is actually not good at parsing. It works best to build dictionary for other more advanced parsing systems.
##################### Original code repositary from https://github.com/LouYu2015/analysis_on_the_story_of_a_stone I automated different parts of code. However, it is slow working with large text data. I am optimizing it further.