python3 ./data_prepared/summarize_corpus.py
python3 ./data_prepared/summarize_facts.py
python3 ./data_prepared/divide_corpus.py
python3 SQA.py
python3 SQA.py --from_checkpoint True
python3 SQA.py --state 1 --from_checkpoint True
python3 SQA.py --state 2 --from_checkpoint True
/checkpoint :
the default folder to save checkpoint of MemNN
/corpus :
/original:
the original datasets
sum_corpus.txt:
default file which collect all dataset in /original folder
testset.txt:
default test dataset
trainset.txt:
default train dataset
/data_prepared :
divide_corpus.py:
divide the /corpus/sum_corpus.txt to /corpus/trainset.txt and /corpus/testset.txt
summarize_corpus.py:
summarize dataset in /corpus/original in /corpus/sum_corpus.txt
summarize_fact.py:
summarize knowledge base in /facts/original in /corpus/sum_facts.txt
/facts :
/original:
the original knowledge base
sum_facts.txt:
default file which collect all knowledge base in /original folder
/intermediate:
save intermediate files generated in training process and used in test/answer process
/MemNN:
MemNN and its components
/transform:
json object which could map mid/IMDBid to labels and map labels to mid/IMDBid
common.py
SQA.py:
main file of the project