Advanced Statistical Language Processing: Reading the Web (10-709)

Homework 3
Due October 15, 2009

Tom Mitchell
Machine Learning Department
School of Computer Science, Carnegie Mellon University

Fall 2009

Read the paper:

Active Learning Literature Survey, B. Settles, Computer Sciences Technical Report 1648, University of Wisconsin–Madison. 2009.

Write out answers to the following questions in a page or two. The point of the exercise is to give you a chance to read about active learning, and come up with your own ideas about how it can apply to a never-ending language learner. The paper author will lead a discussion in class.

1. What are some active learning opportunities in the context of a never-ending language learning (NELL) system?

2. How are these opportunities similar to or different from the problem settings in previous work on active learning? Think about the kinds of queries the learner can ask and how the labels are obtained.

3. Identify some practical issues for active learning in a never-ending learning system like NELL. (Note: don't be limited by issues discussed in the survey, think broadly!) How might we address these issues?