Question_Answering_Models

This repo collects and re-produces models related to domains of question answering and machine reading comprehension

comunity QA

Dataset

WikiQA, TrecQA, InsuranceQA

data preprocess on WikiQA

run preprocess_wiki.ipynb

Pointwise Style

Siamese-NN model

This model is a simple complementation of a Siamese NN QA model with a pointwise way.

train model

python siamese.py --train --model NN

test model

python siamese.py --test --model NN

Siamese-CNN model

This model is a simple complementation of a Siamese CNN QA model with a pointwise way.

train model

python siamese.py --train --model CNN

test model

python siamese.py --test --model CNN

Siamese-RNN model

This model is a simple complementation of a Siamese RNN/LSTM/GRU QA model with a pointwise way.

train model

python siamese.py --train --model RNN

test model

python siamese.py --test --model RNN

note

All these three models above are based on the vanilla siamese structure. You can easily combine these basic deep learning module cells together and build your own models.

Pairwise Style

QACNN

Given a question, a positive answer and a negative answer, this pairwise model can rank two answers with higher ranking in terms of the right answer.

train model

python qacnn.py --train

test model

python qacnn.py --test

Listwise Style

Compare-Aggregate model

To be done

Machine Reading Comprehension

Cloze Style

Dataset

CNN/Daily mail, CBT

GA Reader

To be done

SA Reader

To be done

AoA Reader

To be done

Answer Extraction Style

Dataset

SQuAD, MS MARCO

BiDAF

To be done

Answer Selection Style

Dataset

RACE dataset

Information

For more information, please visit http://skyhigh233.com/blog/2018/04/26/cqa-intro/.