Quizzes

Quiz 0

• Create a Github account if you don't already have one.
• Fork this repository by clicking the Fork button at the right-upper corner.
• Import the forked repository to the IDE (e.g., Eclipse, IntelliJ) as a Java project on your local machine.
• Compile the project with either a command line or your IDE.

mvn compile

• Run the MavenTest.
• Take the screenshot showing the success of your unit test and rename it with your full name (e.g., jinho_choi.png).
• Put the screenshot under your img folder.
• Commit and push your change to your Github repository.

Quiz 1

• Read the post on POSTagger.
• Download the training, development, and evaluation data: wsj-pos.tgz.
• Run POSDevelop and show the difference between:
• Perceptron vs. AdaGrad.
• Averaging vs. non-averaging.
• L2-regularization (extra credit).
• Create a wiki page Part-of-Speech Tagging Results under your repository and report your findings.

Quiz 2

• Run the POSTrain using the following parameters:

  -c src/main/resources/configuration/config_train_pos.xml -t wsj/pos/trn -d wsj/pos/dev -te pos -de pos
  

• Try to build the most accurate model by applying:

• Different learning algorithms using different hyper-parameters.

• Different feature extractors.

• Create a wiki page Part-of-Speech Tagging on WSJ page under your repository and report your findings.

• The top scorers win extra credits.

Quiz 3

• Create unit tests for DEPNode under DEPNodeTest.
• Fill the test method that assesses important methods in DEPNode.
• The objective of this quiz is to be familiarized with the APIs in DEPNode, which is important for the homework #1. There are many functionalities already built into this class so please take a close look at this class.

Quiz 4

• Create a wiki called Word2Vec Explained.
• Explain how the negative sampling works for bag-of-words and skip-grams in terms of our codes in:
• Word2Vec.
• NegativeSampling.

Quiz 5

• Write a method that takes two arrays of NERNode and returns the F1-score of the named entities with exact match. Create a wiki called Named Entity Recognition and put your method in the page as follows.

public double getF1(NERNode[] gold, NERNode[] sys)
{
    return 0; // TODO:
}

Quiz 6

• Write a method that returns the argument candidates of the predicate using the higher-order pruning. Create a wiki page called Semantic Role Labeling and post your answer.

public List<NLPNode> getArgumentCandidateList(NLPNode[] nodes, int predicateID)
{
	// TODO: return argument candidates using the higher-order pruning.
	return null;
}

Quiz 7

• Create three lists of English common nouns, 1) consisting of implications of male (e.g., boy), 2) consisting of implications of female (e.g., girl), and 3) consisting of implications of person (e.g., cousin, tailor). Create a wiki page called Gazetteers and post your lists with a brief explanation of how you created the lists (make sure to indicate the source of your resource).