qa2hypo

Transfer question and answer pairs into assertive sentences.

Quick start

1. Download and install the nltk package that is used in the project to parse parse-trees.
2. Download the Nodebox English Linguistics library and put it in the root directory of the project.
3. Run python ./stanford-corenlp-python/corenlp.py for using the stanford corenlp package for constituency parsing.
4. Call qa2hypo(question, answer, corenlp) in qa2hypo.py to get the transformation in str structure. To optimize for speed in compensation for performance, set argument corenlp as False, and accordingly there is no need for step 2.
5. This module is part of DQA-net. To work with DQA-net, tensorflow installation is required. Common issues include an outdated version of libstdc++6, and the solutions can be found here and here.
6. This module also works with Math Question Answering and its previous version, which is under heavy development at this point, aiming at automating algebraic question answering.