ashg1910/srl

Semantic Role Tagger for LREC 2018

SRL

Semantic Role Tagger for LREC 2018

For testing this code, dummy files have been given in the folder '/finalTraining' and '/argument-classification/allTest/finalTest'

To run the code : Use python version 2.7 or later Libraries needed - sklearn, numpy, gensim, pickle, keras

How to run:

1. -> python prune_sentences.py finalTraining/ #takes out the sentences which don't have pbrole annotation.
2. -> python feature_extraction.py finalTraining/ #when given a propbank data file as input, it creates features of the files in this folder.

In the folder 'wordvecTrain/'

python trainCBOW.py hindi_corpus/ #train a word2Vec model using CBOW

In the folder 'argument-classification/allTest/' #do the same things

3. -> python prune_sentences.py finalTest/

4. -> python testFeaturesExtraction.py finalTest/

5. -> python testVectorizer.py testFeats.txt #formed after step 4.

In the folder 'argument-classification/' -->

6. -> python clVectorizer.py ../trainFeats.txt #trainFeats.py is formed after step 2.

7. -> python crossVal.py #gives out the results after 5 cross validation.

For data set queries, you may contact -> [email protected]