ma-sultan/short-answer-grader

An automatic short answer grader.

short-answer-grader's Issues

No license

Great work, if you don't mind, please include a license (ideally MIT) so that others may build off of it.

error in CoreNlpUtil.py

when I run the code, I have this error in file CoreNlpUtil

left = item[1][0:item[1].rindex("-")]
ValueError: substring not found

when I debugged the code , ' item ' has no - in its string, how to solve this please?

Error while trying to run trainAndApplyGrader.py

After running trainAndApplyGrader.py key error is shown in "if len(parseResult ['sentence']) ==1" in the coreNlpUtil.py file.
Please I need help to solve this error. Urgent help required

missing corenlp.py

Hi,I'm a student who is learning NLP recently.It's my honor to learn your demo.But where is the corenlp.py?

Porting code to Python 3

I tried porting the code to python 3.
2to3 was helpful in automating the whole process.

Also I had to do the same with the standford-corenlp-python library.

But the code fails in the socket connections:

'''
Traceback (most recent call last):
File "pretrainedGrader.py", line 66, in
score = grade(question, ref_answer, student_response)
File "pretrainedGrader.py", line 39, in grade
feature_vector = extract_features(question, ref_answer, student_response)
File "pretrainedGrader.py", line 12, in extract_features
sts_alignment(ref_answer, student_response)
File "/home/deebuls/Documents/digi-klausur/short-answer-grader/featureExtraction.py", line 50, in sts_alignment
sentence1_parse_result = parseText(sentence1)
File "/home/deebuls/Documents/digi-klausur/short-answer-grader/coreNlpUtil.py", line 20, in parseText
parseResult = nlp.parse(sentences)
File "/home/deebuls/Documents/digi-klausur/short-answer-grader/coreNlpUtil.py", line 14, in parse
return json.loads(self.server.parse(text))
File "/home/deebuls/Documents/digi-klausur/short-answer-grader/jsonrpc.py", line 934, in call
return self.__req(self.__name, args, kwargs)
File "/home/deebuls/Documents/digi-klausur/short-answer-grader/jsonrpc.py", line 907, in __req
resp = self.__data_serializer.loads_response( resp_str )
File "/home/deebuls/Documents/digi-klausur/short-answer-grader/jsonrpc.py", line 598, in loads_response
raise RPCInvalidRPC("""Invalid Response, "jsonrpc" must be a string.""")
jsonrpc.RPCInvalidRPC: <RPCFault -32600: 'Invalid Request.' ('Invalid Response, "jsonrpc" must be a string.')>

'''

What i understand is Python has introduced new format called Byte.
So string in python2 is different from python3.

I have commited the converted python 3 files in my branch here : https://github.com/deebuls/short-answer-grader/tree/python3_port

Also the python 3 port of stanford-corenlp-python is here : https://github.com/deebuls/stanford-corenlp-python/tree/port_python3

I am currently looking into the problem. Maybe you can provide some insights.

Dataset for testing

Thanks for releasing the code with the paper. Nice contribution to reproducible research.

I wanted to know were you able to get the Mohler dataset . I looked online couldnt find.