hankcs
hankcs
It's a variant. As long as you use the same formula in both building and searching phases, it is right. https://github.com/hankcs/pyhanlp/blob/49202d2210d97b075e35c90aa20b9cbe3b31f413/tests/book/ch02/dat.py#L32
No, although I wrote this after the paper you mentioned, I did not read it and I can't tell if the algorithms are the same.
No, but I wrote a blog post on it:https://www.hankcs.com/program/algorithm/aho-corasick-double-array-trie.html You may need a translator.
It's a good idea but I don't have time now. I'm pretty sure there is a go implementation somewhere on GitHub.
No, not safe at all. How could it be safe when its states are getting updated during building?
每个单词有两个feature，一个是Word Embedding，一个是Case Embedding，see：https://github.com/hankcs/CS224n/blob/master/assignment3/data_util.py#L45
没问题，等周末有空就上传。
They belong to different matrices. Input vectors are from lookup table and output vectors are from hidden matrix. You may need to go over the lecture video.
https://github.com/hankcs/HanLP/wiki/%E8%A1%8D%E7%94%9F%E9%A1%B9%E7%9B%AE#elasticsearch-analysis-hanlpkennfalcon
Yes.