Beefy Boxes and Bandwidth Generously Provided by pair Networks
"be consistent"
 
PerlMonks  

Re^2: RFC: How to Write a Great Thread

by jhourcle (Prior)
on Mar 05, 2009 at 17:56 UTC ( #748619=note: print w/ replies, xml ) Need Help??


in reply to Re: RFC: How to Write a Great Thread
in thread RFC: How to Write a Great Thread

The problem with keywords is that what you think of the problem is might not be what everyone else thinks of it as.

For a while now, I've been trying to figure out how I can do 'server push' with an AJAX request. Finally, a couple of weeks ago, I found that there's a whole website that talks about the general concept, but they call it 'Comet'.

If you're looking for documents about AJAX from _before_ it was called AJAX (and I'm not talking about sports teams, cars, or cleaning products), they're not going to have that keyword associated. Once in a while, terms used to explain concepts diverge or converge, or just differ between communities.

When you're dealing with search systems, you typically measure its performance in terms of 'precision' (you only get back relevant results; bad precision == information overload) and 'recall' (you get back _all_ of the relevant results). Adding too many keywords (especially if the keyword is polysemous) helps with recall and the whole 'search engine optimization' efforts to boost your ranking, but it harms precision if you're only peripherally related to the extra keywords that you insert.

I'd love for search engines to come up with a 'context' concept, so I can search for 'apple' in a food context, or 'map' in a perl context. (these aren't the best examples ... the issue comes in ranking ... I'd NOT want the context to be used for sorting of the results ... so 'map' is the important criteria and 'perl' is only used to filter down the list to remove non-perl contexts) ... I just wish I could remember some of the times when I've actually run into this problem.

(sorry ... my library school classes on classification theory and cataloging are slowly floating to the top ... and this morning's meeting on how to define certain concepts used in classification for a metadata standard)


Comment on Re^2: RFC: How to Write a Great Thread
Re^3: RFC: How to Write a Great Thread
by zentara (Archbishop) on Mar 05, 2009 at 20:16 UTC
    ++, got me thinking. Searching with a "context" is a good idea.....but how to define context from a word in random phrases? "apple of my eye", "apple tree", "Apple Computers", "road apple", etc. :-)

    Google is probably working on it now.


    I'm not really a human, but I play one on earth My Petition to the Great Cosmic Conciousness

      Actually, I'm pretty sure Google already does that, in a way. If you're logged in to one of Google's systems (like Gmail), it knows what you've searched for before, and uses those data to compute your search results, so that if you've done lots of searches on computers and such, computer results get a higher weight. IIRC they do some weighting based on location as well.

      Incidentally, I would really, really like to know more about how Google's algorithms work. I'm pretty sure those methods being known would be a great boon to research in those areas of computer science.

Log In?
Username:
Password:

What's my password?
Create A New User
Node Status?
node history
Node Type: note [id://748619]
help
Chatterbox?
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others contemplating the Monastery: (6)
As of 2014-10-24 23:29 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    For retirement, I am banking on:










    Results (138 votes), past polls