Test and Research Question Classification

[schwittlick]

when the input is a question, the reply shouldn't begin with what the input was.

[schwittlick]

http://cogcomp.cs.illinois.edu/Data/QA/QC/

[schwittlick]

some text parser that seems worth checking out. written in java though:

http://nlp.stanford.edu/software/lex-parser.shtml here are python implementations of that (jython/jpype):

https://github.com/vpekar/stanford-parser-in-jython http://projects.csail.mit.edu/spatial/Stanford_Parser

[schwittlick]

if (sentence ends with question mark || begins with 'how'/'why'):
    if begins with 'how':
        answer with seed 'it is'
    if begins with 'why':
        answer with seed 'because'

[schwittlick]

some more question classification/detection happening here https://github.com/mrzl/ECO/issues/160

[schwittlick]

dataset containing 400.000 similar question pairs: 800.000 questions that could be used for training a CNN to detect questions automatically: https://data.quora.com/First-Quora-Dataset-Release-Question-Pairs