Find two sentences licensed by the grammar on slide 5 of 11/4, such that the longer sentence has a higher probability than the shorter sentence. Show the trees assigned to each sentence by the grammar, with a probability on each node, and the probability of each tree.
[NB - This grammar is just like the one in Fig 12.1 (p.449), except that the probability for Proper-Noun -> Denver has been corrected to .60.]
Step through the corrected version of the pseudocode for
probalisitic CKY (see slide 11 of 11/4)
to see how it would parse the following sentence, given this grammar:
Kim adores snow in Oslo.
Your write-up should include two things:
We recommend using a format like the following for displaying the chart: (NB - This figure is intended solely to illustrate the display method and bears no ressemblance to the chart for the sentence you are considering.)
1 | 2 | 3 | |
1 | Nom: .004 N: .08 | S: .0008 | |
2 | VP: .0001 V: .09 | VP: .003 | |
3 | NP: .01 Nom: .05 N: .08 |
Hint: The counter m in the second to inner-most loop is what determines the order in which the two possible VPs will be considered.
Do problems 18.3 and 18.4 on p.716 of the textbook.