Insurance QA data formatted as Python objects and pickled.
Clone locally
git clone https://github.com/codekansas/insurance_qa_python.git
cd insurance_qa_python
pwd # where files are stored
Load a file in Python
import pickle
def load(file_name):
return pickle.load(open(os.path.join(path, file_name), 'rb'))
About files:
vocabulary
:dict
object of(word index <int> -> word <str>)
relationshipsanswers
:dict
object of(answer index <int> -> word indices <list of ints>)
relationshipstrain
:list
ofdict
(onedict
per entry), where eachdict
has:question
: the word indices for the questionanswers
: the answer indices for each of the question's ground truth
dev / test1 / test2
:list
ofdict
(onedict
per entry), where eachdict
has:question
: the word indices for the questiongood
: the ground truthbad
: the other answers from the dataset
Applying Deep Learning to Answer Selection: A Study and An Open Task
Minwei Feng, Bing Xiang, Michael R. Glass, Lidan Wang, Bowen Zhou ASRU 2015