• sentiment.zip
  • Funcs.java: Helpful utility functions for processing input data
  • Histogram.java: Represents a histogram. You need to implement getPluralityWinner() as part of a kNN implementation.
  • LabeledWords.java
  • Sentence.java
  • TestResult.java
  • TextLearner.java
  • NaiveBayes.java
  • BayesTest1.java
  • BayesTester.java
  • SimpleTester.java
  • DataSetReader.java
• Create a Naive Bayes Classifier for the following task:
  • The given collection of sentences contains labels of positive and negative sentiment.
  • The goal of the classifier is to predict the sentiment of new sentences.
• Create a k-nearest-neighbor classifier for this task, and compare its performance with Naive Bayes.
• Also find another labeled document collection on the Internet, and build two classifiers for this collection as well.
• Intermediate presentation (2/13)
  • What other document collection will you classify?
  • Present concrete results from one of your two classifiers.
• Final presentation and report (2/18)
  • Relative performance of Naive Bayes and kNN
  • Mention any algorithmic/parameter variations
  • Relative difficulty of the document collections

Grading criteria