I590/N564/N364 QUIZ #4 - 31 March 2006	(PERFECT SCORE: 60+10 POINTS)

                                                 Name _____________________________
													
		
	1. We talked a lot awhile back about the dangers of ambiguous or misleading
	terminology. The problem with it can be thought of in terms of IR evaluation.
	
5>	(a) Is referring to the "number of voices" a synthesizer can handle likely
	to lead to precision (false positive) problems, recall (false negative) problems,
	or both? Why? 

	BOTH. FALSE NEGATIVES BECAUSE VERY OFTEN THE DESIRED INFORMATION WILL BE
	EXPRESSED IN DIFFERENT WORDS, SO A SEARCH FOR THAT PHRASE WON'T FIND IT.
	FALSE POSITIVES BECAUSE "NUMBER OF VOICES" CAN EASILY MEAN SOMETHING VERY
	DIFFERENT FROM WHAT'S INTENDED, SO A SEARCH WILL FIND A LOT OF IRRELEVENT
	ITEMS. (TRY IT WITH GOOGLE!)


10>	(b) I showed a Venn diagram illustrating the fact that precision and recall
	are fundamental concepts that apply to _any_ Boolean (yes/no) classification.
	The diagram had two circles and it divided the rectangle into four subareas.
	Draw such a diagram in the box below. Label both circles, and label the "relevant
	and not retrieved" subarea.
	
	        --------------------------------------------------------
	       |                                                        |
	       |                                                        |
	       |                                                        |
	       |                                                        |
	       |                                                        |
	       |                                                        |
	       |                                                        |
	       |                                                        |
	       |                                                        |
	       |                                                        |
	        --------------------------------------------------------
	

	
10>	2. We talked about three ways of finding music you might be interested in,
	and two general situations: you want it _now_, or you want to keep finding
	it whenever it becomes available. Name the three ways, and give specific
	examples (if possible, existing systems) for two of them.
	
	* BY CONTENT
	
	* BY METADATA (BIBLIOGRAPHIC INFORMATION)
	
	* BY COLLABORATION (INFORMATION KNOWN TO ANOTHER ENTITY)

	
	
10>	3. What's the difference between an _information need_ and a _query_?

	AN INFORMATION NEED IS ABSTRACT (STRICTLY SPEAKING, IT'S AN IDEA OR
	SET OF IDEAS); A QUERY IS THE CONCRETE EXPRESSION OF AN INFORMATION NEED
	IN A FORM SOME COMPUTER SYSTEM CAN USE.


	4. Think of a musical idea in some song or other piece of music you know,
	a musical idea you can describe precisely in some way.
	(a) What piece of music is it in? What genre? (This is just to help me judge
	your answers to the other parts of this question.)
	
	

	
10>	(b) Describe that musical idea as precisely as you can, using any marks you
	can put on this sheet of paper (or another sheet, if you need more space):
	words, one or more diagrams, guitar tab, music notation, etc. Use your
	imagination!
	



	
10>	(b) Do you think it might be possible to describe the above musical idea to a
	computer system some day? Why or why not?
	
	



5>	5. (a) What's wrong with sequential searching for use in "real-world" situations,
	where you might have millions of documents (e.g., tracks) to search?
	
	IT'S TOO INEFFICIENT.
		
	

	
10>	(b) EXTRA CREDIT. We discussed two alternatives to sequential searching. Name one
	and describe how it works as accurately as you can.
	
	* INDEXING (WITH INVERTED LISTS): LIKE THE INDEX OF A BOOK
	
	* SIGNATURES