Homework 4

Named Entity Recognition

Your task is to develop a named entity recognizer that takes a sentence and recognizes each named entity using the BILOU notation:

• Download the CoNLL'03 shared task dataset.
• Develop a model by converting features into sparse vectors.
• Develop another model by converting features into dense vectors using word embeddings (e.g., google news).
• Optimize your models using different features and optimizers.
• Run the most optimized models on the evaluation set and save their outputs to conll03.eng.tst.sparse.tsv and conll03.eng.tst.dense.tsv.
• Write a report describing your approaches, results, and analysis in latex using the ACL template.

Submission

• Compress your code, output, and report into hw4.zip and submit to: https://canvas.emory.edu/courses/29596/assignments/34120