PHP Code for Bayesian Probability Algorithm (possibly C++)

Ditutup Dipasang Jan 9, 2011 Dibayar saat pengiriman
Ditutup Dibayar saat pengiriman

I'm looking for someone to write PHP code for a Bayesian based AI script.

The resulting code should be able to:

* Output to command line (with debug options, etc.)

* Output to html format (with tables, input forms, etc.)

* Use a MySQL database (VARCHAR column) for both learning and guessing options (the html form could be a single form with a rado button for learn/guess)

This AI should be capable of training based on :

* Text file uploaded to the form.

* Textarea pasted by a user into the form

* DB Entry passed to it from the command line.

But not just for a single outcome, such as the way spam filters are built.

In other words, I need probabilities of inputted text that they match each defined category, not just a yes or no.

As a a basic example of what I'm looking for:

I would input a category followed by text to train the AI so that it knows that category.

Example:

Category = Dogs

Text file (or database or html form textarea input) = Articles about dogs

(sample run: #php [login to view URL] --learn --infile=[login to view URL])

Category = Cats

Text file (or database or html form textarea input) = Articles about cats

(sample run: #php [login to view URL] --learn --infile=[login to view URL])

Then, when a new text is inputted, say a sentence, or a text column in a DB or from the same form's textarea, I would like it to tell me the likelihood of that entry matching each of the learned categories (cats and dogs in this example).

Input: A third/new article about Household pets:

(

sample run: #php [login to view URL] --guess --infile=[login to view URL]

Output:

23% likely that it's about dogs

4% likely that it's about cats

)

This input/output functionality should be available via web interface as well.

This project should be able to scale to very high numbers.

It may eventually be analyzing up to 50k lines of text per SECOND (for guess operations only, not training)

- note, however, that each line will never be larger than 1k in length.

For this reason, I would be willing to entertain C/C++ code, but I'm very rusty with those so I would prefer PHP.

(I realize Perl is the better choice here, but I do not have luxury of using it unfortunately)

There's a VERY good tutorial about how to build a php bayes script here, but it's a bit above my head - so I need your help!

[login to view URL]

Pemrograman C Pemrograman C++ MySQL PHP

ID Proyek: #907961

Tentang proyek

8 proposal Proyek online Aktif Feb 13, 2011