CSCI-GA.2590 - Natural Language Processing -- Spring 2013 -- Prof. Grishman

Assignment #7

March 26, 2010

Set up your Jet system to run the appointment extraction task.  You will need

Put the first four files in the Jet data directory, and the properties file in the props directory.  A guide to Appointment patterns summarizes the new features of these pattern files, most of which were described in the lecture notes.

1. [2 points] Run Jet using the appointmentPattern file and the appointments text file.  Display the output (just the lines with the events found). Note:  these will appear in the DOS or shell window, not the Jet console.   Identify at least five errors in the output, and the cause of the errors.

2. [2 points] Modify the chunkPartternPA and the appointmentPattern files to correct at least three of these errors.  (Improvements you have previously made to chunkPatterns may be used.)

3. [1 point] Run the initial pattern sets and your improvements on the Wall Street Journal newswire of January 12, 1996 and report the results of your improvements.

As before, a small extra credit will be awarded for particularly effective pattern improvements.

Due April 2nd.