Monday, February 13, 2012

Tactic: Scoping

I've written about scoping before, but it's worth revisiting with a little amplification.  


Scoping just means limiting the space of things you're searching over.  A great example of this is the way a search on Maps.Google.com limits the search to the part of the map that's visible.  


Here, I start with the map I see after doing the search for [ Plano, TX ]... 
When I do a search for [ coffee ], my search is limited to the boundaries of the map--that is, it's scoped to the visible map and not all of the shops in the Google index.
  




The great thing about this kind of scoping is that you can see the effects of changing the scope!  If you zoom in, the results in the periphery go away.  If you drag the map side-to-side, the results will change depending on what's in the scope.  


In essence, all scoping is like this.  You start a search with the scope of the index you're searching.  For instance, you could scope your search to the contents of JSTOR, if you're searching their index.  Or you could scope your search by using an operator like filetype: or site:  Big point:  You're not going to find things that are out-of-scope.  


Why do you care about this?  Two reasons:  


(1) Over-scoping:  if you set your scope too tightly, you can waste a huge amount of time searching for something that's really not there.  In my studies (and in my own searches), I see this all the time.  Someone scopes a search to a piece of the internet that doesn't have an answer.  (See the upcoming post about "Know when to quit.")   For instance, if your search path when looking for an article from NASA about the flight of Apollo 13, you might do a search like this:  [ NASA Apollo ] then scope it down a little more by adding [ NASA Apollo site:.gov ] and then a little more [ NASA Apollo site:.giv filetype:txt ] and lastly, add yet another scoping tool by using minus:  [ NASA Apollo site:.gov filetype:txt -release
You can see what's happening at each step... you're focusing the search onto a smaller and smaller piece of the index.  The first query has around 2.5M results; the second query reduces that to 1.39M, etc.  

But suppose the thing you're looking at isn't at NASA.gov at all?  What it were on the NYTimes.com site?  Then you'd be overly scoped and stuck searching in a part of the index that has no chance of containing your  desired search target.  




(2) Focus your search:  Typically, scoping reduces the number of possible hits given your query.  You might start with millions of hits, and if you go down to the right scope, then you can do a query that would be meaningless in the wild and wooly larger scope, but very meaningful in the smaller scope.  


In an earlier post (SearchResearch of Oct 13, 2010) I pointed out how limiting your search to the scope of an encyclopedic book can give you wonderful power.  In that example I showed how using Book search to search within the text of David Kahn's "Codebreakers" allows you to search for the single word "cliche" and find the answer.  Obviously, if your search isn't scoped to just the contents of that book, then a single-word query like that wouldn't work very well.  


Ways to scope:  I've given you a few of the ways I do scoping.  


A. Select a corpus that you want to search (such as Google Images, Google Patents, Google Books, etc.)  The more corpora you know, the better off you are.  (Especially if they're tuned to the kind of thing you're looking for.  It's smart to look for 3D models of objects in the 3D-Warehouse rather than, say, Patents.)  


B.  Use advanced operators when you can.  My most-used operator is probably site: followed closely by filetype: -- both of these scope your search tightly.  (But be aware, if you can't find it within the scope, consider stepping back a few scopes and rethinking your strategy.)    Remember that the - (minus) operator also acts to scope your search by removing some of the hits.  


C. Books can be a scope too.  On more than a few occasions I've found something by doing a full-text search inside of a book text.  Remember that Google Books has a ton of books, as does Amazon (usually more recent book texts).  There are other repositories worth remembering as well.


D.  Maps are implicitly scoped.  Zoom into the region you want to search, then do a search on that map.  (Useful for businesses and places.)  


More tactics and strategies to come, but this will get you started on scoping! 


Search on!  (And always within a scope that you understand.)  

No comments:

Post a Comment