Date: Tue, 16 Jul 2002 10:56:38 -0500 (CDT)

Subject: developing a critical reading skill


Again, this is a very important skill that nobody usually
teaches you in school, but it is very crucial in my opinion,
so read the following *very* carefully and make it a habit.


1. Evaluation of a paper

When you read a paper, it is hard to judge whether it is good
or bad and because the authors usually highlight their strengths,
not their weaknesses, it is hard to see if the paper is significant
or not, and what might be the shortcomings of the papers.

One way to evaluate this is to totally understand the paper and
think about it over months and come to a conclusion yourself.
By that time you would have gained a lot of knowledge about that
topic and can make an intelligent evaluation of the paper.

However, this is TOO time consuming. There are thousands of papers
publised everyday and you cannot possibly do this. An easier
way is to;

	1. Look for the author's own evaluation of the work.
		- see what they think their shortcomings are
		  and how important their work is, what is new,
		  what are their contributions.

	2. Look at papers published later that cite the given paper.
		- see how they compare the previous paper to others
		  and to their own.

By looking at other people's evaluation of a certain paper, you can
quickly see what are the things the original authors did not see.


2. Evaluation of books and gaining a broader perspective quickly

A similar strategy works well for single author books. When you find
a book interesting, read the book reviews first. Go to amazon.com, etc.,
and also there are independent reviewers. These two are the ones
that I like particularly:
	
	http://dannyreviews.com/
	http://www.santafe.edu/~shalizi/reviews/

One good thing about book reviews is that you can often gain a good
broad perspective on a given topic or the content of a book by just
reading the reviews. This is a really good way to quickly catch up
with a certain topic, and find out what are the relevant issues.
Also, you can decide to read the whole book later if you find
the reviews to be interesting.

Another source of good book reviews (for the more technically-oriented)
are Journals. For example, Science, Nature, and many other journals
have book reviews of books in their specialized areas. These reviews are
written by top figures in the field and they also provide great
perspective and intuition.


3. Gaining perspective

There are other good ways to gain perspective on a topic. In journals,
there are articles called "Reviews" or "Minireviews", which are basically
a paper summarizing the progress in the area. The reviews usually 
list all relevant work and try to draw a big picture. For example, the
Abbott and Simoncelli paper on natural scene statistics is one good
example. 

Instead of digging into papers that are highly technical, it is often
a very good idea to read these reviews first.

There are journals dedicated to such reviews: Annual reviews of XXXXX.

	http://www.annualreviews.org/

Also, to repeat from the previous email, Edited volumes have good
introductory material that are sort of a small review of the area.

These are good places to begin.

Another good source are PhD dissertations. Especially the introduction,
related work, and discussion. Dissertations are usually written to be
self-contained, so even if you don't understand the technical part,
it helps to read the perspective.


4. Selecting a topic to work on

You will often see some interesting ideas in the discussion section
of a paper or whatever. They usually talk about future work, etc.
Try to see if there's anything that is interesting to you that you 
feel confident that you can actually get it to work.

5. Organizing things

Also, make notes when reading a paper. I usually have a huge text file
containing my ideas and excerpts from papers. This way, you know where
to find your stuff, and it is easy to search. More on this later.


Yoonsuck Choe
choe@tamu.edu