Homework #4: Validity vs. Reliability

Paper

Christopher D. Manner & Hinrich Schütze (1999), Foundations of Statistical Natural Language Processing, MIT Press, Chapter 5: Collocations.

Sections you need to read to finish this homework

• Opening introduction
• §5.1: Frequency
• §5.5: The Notion of Collocation

Ouestion

Story

This chapter discusses a variety of computational approaches to finding collocation. Having read this, two motivated students, John and Monica, are eager to experiment with these ideas:

• John wants to experiment with the approach mentioned in Section 5.1, using frequency and part-of-speech filtering techniques.
• Monica chooses to experiment with another one outlined in Section 5.5, using the 3 factors: non-compositionality, non-substitutability, and non-modifiability.

Both students believe in corpus-oriented paradigm, of course.

Show time!

Now, it’s your job to

1. evaluate (all right, alright, predict) the quality of their researches in terms of validity and reliability.
2. identify potential obstacles of each approach and, if possible, give useful suggestions to help improve the quality.

Important Dates

Announce: 2004-03-30    13:30

Due: 2004-04-06    13:30