HW #4: MORPHOLOGY AND FINITE-STATE TRANSDUCERS

(revised)

1. J&M, Ch. 3, p. 89: #3.6: Implement the Soundex algorithm by drawing an approprate TN.

2. Here is the stemming algorithm for `-ly' words from Smith, George W. (1991), Computers and Human Language (New York: Oxford University Press): 26, Fig. 4:
   IF the word ends -ly
   THEN remove the -ly and
   
   IF the remaining string does not end -i
   THEN IF string doesn't match a stem
   
   THEN add -le and
   
   IF string matches a stem
   THEN yes!
   ELSE end
   
   ELSE yes!
   
   ELSE change -i to y and
   
   IF string does not match a stem,
   THEN end
   ELSE yes!
   
   ELSE IF string matches a stem
   
   THEN yes!
   ELSE end

   • Test it on the following words:
     mainly, easily, ably, rely, belly, oily, early, fly

   • Describe any problems you find, and write a new algorithm that fixes them.

3. Play with Englex to find its capabilities and limitations. Discuss your findings. Hand in sample runs to justify your conclusions.

DUE: MONDAY, FEBRUARY 14