CS 1671
Chapter 19 Study Guide
======================

**F** Marks a possibility for the final.

The questions prefaced by Q have solutions in the *studyguideSol.txt file.

Q:**F**  Look up "sewer (noun)", and "plane (noun)".  What do you think the homonyms are?

Find in WordNet other examples of the relationships in 19.2 and 19.3. Click on the "S" to see the relationships to other synsets.  Click on things, drawing a graph, until you understand the structure of WordNet.  Note words with synsets that have different hyperny chains (as in Figure 19.4).  If something confuses you, please write down the particular word/POS, and ask me.

You do not need to memorize the names of the lexical relationships, except for "hypernym" and "hyponym" 

Section 3.4 of (resnikYarowsky2000.pdf) Resnik & Yarowsky (2000)

   Given parallel corpora between English and the other languages in Table 4, show how we can create English training data for word-sense disambiguation without requiring manual annotation.
  
   **F** For example, which German words give us training data for sense 1 of "interest"?   Why do "provecho" and "benefio" give us training data for sense 5 of "interest", but "interes" does not? (These were   covered in Lecture.)

   **F** Which sense distinction is dropped?  Which sense is split?  Which  sense is eliminated? (These were covered in lecture.)

    Q:**F**  What training data for the senses of "interest" could we get from Table 4 and Japanese-English parallel corpora?

Make sure you can do the following (we did a similar example in lecture):
    Give two syntax trees for "I ate somewhere close to work"
    For each of your trees, give the semantic roles for "ate" (you can call them "AGENT" and "THEME")
    Give a reasonable selectional restriction for the THEME of ate.
    If a system were able to recognize that "somewhere close to work" does not meet the selectional restriction, it could use this formation to determine which syntax/semantic-role analysis is more probable

Q: Now consider "I ate Spaghetti with meatballs."  
   Give two syntax trees for this sentence.
   For each of your trees, give the semantic roles for "ate" (you can call them "AGENT", "THEME" and "INSTRUMENT")
   Give "reasonable" selectional restrictions for the INSTRUMENT of "ate" 
   Again, if the system were able to recognize that "with meatballs" does not meet the selectional restriction, it could use thisinformation to determine which syntax/semantic role analysis is possible

**F** The above two, but with different (but similar)  examples