A review for a conference or a journal normally addresses two goals:
(1) deciding whether or not a paper is good enough to accept, (2)
offering constructive criticism to the author.  For our class, we'll
skip item (1) and just focus on the constructive criticism part.
Below is a reviewing form template you can use for your comments.  It
is a slightly modified version of the 2002 SIGGRAPH papers reviewing
form.



 
 1.  Briefly describe the paper and its contribution. 

 
 2.  Is the exposition clear?  How could it be improved?


 3.  Are the references adequate?  List any references that 
     are needed. Cite specific publications or public disclosures
     of techniques.


 4.  Could the work be reproduced by a skilled
     graduate student?


 5.  Are limitations and drawbacks of the work
     adequately discussed?



 6.  Does the paper discuss the following items? How well?  
     (1) the comparison with alternative
     methods in such a way (e.g., with real or re-used data) as to 
     allow rigorous evalution; (2) the degree to which human 
     intervention is involved in generating the result (i.e., is this 
     an automatic method, or a human-coached method?); and (3) the 
     "brittleness" of the results, i.e., will the method work for ANY 
     data, or only for the models shown, or only for models in some 
     class? Note that the answers to parts 2 and 3 can be "lots of 
     intervention" and "only works for this well-described class of 
     models" and the paper may still be excellent!


 7.  How could the author improve the paper?  Feel free to give
     constructive comments at whatever levels you'd like--from
     wording corrections to paper organization to research
     assumptions.