RE: [Classifier4j-devel] Training and Classifying

[Neil G. <np...@nt...>]

Hi Nick,

Thanks for the reply, however there is still a problem.

The following outlines how I am trying to access the data I have trained
in the "word_probability" table.  I wish to classify the String
'contents' against the data in the "word_probability" table:

	DriverMangerJDBCConnectionManager cm = new
DriverMangerJDBCConnectionManager(connString, user, pw);

	JDBCWordsDataSource wds = new JDBCWordsDataSource(cm);
	
	ITokenizer tokenizer = new DefaultTokenizer();

	IClassifier classifier = new BayesianClassifier(wds, tokenizer);

	classifier.classify(contents);


I get a system error message saying:
net.sf.classifier4J.bayesian.WordsDataSourceException: Problem creating
table

and

Caused by: java.sql.SQLException: ERROR:  Relation 'word_probability'
already exists

It seems as if the JDBCWordsDataSource constructor is always trying to
create a table, when I just want to be able to read from the existing
table!

I am using a Postgres SQL database.

I would very much appreciate you providing some code which enables me to
read from the "word_probability" table and then use the standard methods
to classify an unknown string.

Regards


Neil



> -----Original Message-----
> From: cla...@li...
[mailto:classifier4j-
> dev...@li...] On Behalf Of Nick Lothian
> Sent: 18 May 2004 00:11
> To: 'cla...@li...'
> Subject: RE: [Classifier4j-devel] Training and Classifying
> 
> >
> > Hi Nick,
> >
> > The problem I am having is as follows:
> >
> > I wish to classify an unknown article against trained data,
> > which is in
> > the "word_probability" table.
> >
> > So, to do this I must use BayesianClassifier as follows:
> >
> > 	1. Create an instance of JDBCWordsDataSource
> > 	2. Create a new instance of BayesianClassifier, passing the
> > JDBCWordsDataSource to the constructor
> > 	3. Call BayesianClassifier.classify(String)
> >
> > However, by creating an instance of JDBCWordsDataSource (1), the
> > constructor for this method creates a new table.
> > ------ which I do not want because I will have to delete the table
> > containing all my trained data.
> >
> 
> Well that would be bad!
> 
> Fortunatly you are incorrect about the constructor:
JDBCWordsDataSource
> detects is the table already exists, and if the table does exist it
> doesn't
> create it - see
>
<http://classifier4j.sourceforge.net/xref/net/sf/classifier4J/bayesian/J
DB
> CW
> ordsDataSource.html#235>
> 
> (Note that there is a bug in the 0.5 version of Classifer4J here if
you
> are
> using MySQL on Unix - the table names are case sensitive. Get the CVS
> version).
> 
> If you are actually seeing this behaviour (the table being dropped)
could
> you please give more details about the database you are using, etc.
> 
> 
> > What I want is to be able to retrieve my trained data (in a
> > JDBCWordsDataSource??), then classify the new article against
> > this data.
> >
> > How can I do this?  I would very much appreciate some sample code.
> >
> 
> It sounds like you are on the right track. Let me know if you try this
and
> it doesn't work.
> 
> Nick
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: SourceForge.net Broadband
> Sign-up now for SourceForge Broadband and get the fastest
> 6.0/768 connection for only $19.95/mo for the first 3 months!
> http://ads.osdn.com/?ad_id=2562&alloc_id=6184&op=click
> _______________________________________________
> Classifier4j-devel mailing list
> Cla...@li...
> https://lists.sourceforge.net/lists/listinfo/classifier4j-devel