emory-courses/cs571

Natural Language Processing.

CS571: Natural Language Processing

This is a graduate-level course focusing on topics in Natural Language Processing (NLP) such as embedding encoding, text classification, sequence tagging, semantic parsing, coreference resolution, and machine comprehension. Advanced topics in deep learning with various neural network models in applications to NLP are also discussed.

• Syllabus
• Schedule
• Paper presentations: 2021, 2019
• Final projects: 2021, 2019, 2017, 2015

cs571's Issues

pics

Put model inside SGD

Would it make sense to put the model inside SGD and pass in a new weight vector since every SGD must have a model?

pics

Dead link and question.

Hi,

It seems the links to reader and structure in homework5 instructions do not work. Which reader and structure python files is being referred to? Do we need to use reader.py structure.py and dependency parser.py?

Thanks,
Mani

Proposal Template

Hello professor Choi,

Do we use the other template for the proposal? If I remembered correctly, you mentioned a template in the class. Or we just use the ACL format as our template for the proposal?

Question about test cases for dependency parser

Hi everyone,

Does anyone have another test case( a sentence with the same format as the gold tsv file), along with correct output, that one could use to make sure the dependency parser is working fine, besides the one provided in the homework instructions( preferably longer sentences than the one provided)?

I tried to put the long example used in lectures(attached)
test.tsv.zip

into the format of the tsv file and test it, but I can't seem to get it working, and I keep getting error. I assume only word index and head index columns are important.

Thanks,
Mani

Ideas for dependency parsing

Anyone wants to try structural regularization on dependency parsing?
http://papers.nips.cc/paper/5563-structure-regularization-for-structured-prediction.pdf

[hw4] Is `pos-tagger.py` executable?

Hello,

When I exec the command below:

$ python pos-tagger.py data/conll03.eng.trn.gold.tsv data/conll03.eng.dev.gold.tsv

It shows this result forever:

>> Reading: data/conll03.eng.trn.gold.tsv

Do we need to modify something?