My suggestion is that you either make your proposal more about the algorithm
itself or more about search engine application as seen by the user.

As it stands, your proposal doesn't sound like you know much about either
one of these and are making serious decisions about implementation without
learning more about the factors that would drive these decisions.

It would also help if you have a friend who speaks English well help you
edit your proposal to make it easier to understand and to make sure it says
what you mean to say.  Many readers of your proposal will try to compensate
for the difficulty in communication, but if you can make it easier for them,
it would help very much.

As an example of what I mean by the algorithm based proposal, you could
simply say that you would like to implement SVM as part of the mahout
project, especially optimized for processing text data and relevance
feedback.  

For the user centered kind of proposal, you could say what the user problem
is that you are trying to help with and then describe how your
implementation will help with that.

For a proposal to be successful as an implementation using map-reduce, you
should say why it is important to use parallel processing.  Note that it is
NOT usually important to use a large parallel cluster for processing
relevance feedback because there are only a few training examples in these
cases.  It is also not usually feasible to use SVM in a large web index
because there is no training data and because SVM training cost goes up
dramatically with the size of the problem.


On 3/29/08 11:14 AM, "Marko Novakovic" <[EMAIL PROTECTED]> wrote:

>>>>>>> I noted that the most usable solution for
>> search
>>>>>>> engines is Support Vector Machine.
>>>>>>> The best solution for determination relevant
>>>> page
>>>>>>> ranking for user based search result is SVM.

Reply via email to