Develop Reference

library for text classification in java

I have a set of categorized text files. I want to categorize another large set of text files to use in my research. Is there a good way to compare them?
I think SVM based methods are useful but is there a simple and documented library for using such algorithms?

I don't know much about SVM, but LingPipe might be really helpful for you. The link is a tutorial specifically about categorization of documents (automatic or guided).
Also, look into the inter-related search products Lucene (a search library), Solr (search server app), and Carrot2 (for 'clustering' search results). There should be some interesting work in that space for you.

Mallet is another awesome library to look into. It has good commandline tools to help you get started and a Java API once you start getting into integrating it with the rest of your system.

Common Java based PDF construction libraries/tools?

I'm looking for 2 or 3 of the most common/industry wide used libraries for the Java platform for the creation of PDFs on the fly.
The one requirement I'm focusing on is the ability to use specific formatting such as page layout and font sizes and typefaces (this will be a dynamically created legal document with frustratingly specific type standards).
I'm not actually going to be the one implementing this (I'm not a Java developer), but am trying to get the ball running and need to pass along some things to have our dev team to start investigating.
I'm investigating iText at the moment, which seems to be a well established option. I'm not yet sure how robust/flexible the templating abilities are, though.
EDIT: I just realized that there's probably no one 'right' answer for this question so maybe htis is better as part of the Wiki.

iText is probably the best all around free tool.
PDFLib is another choice if you are willing to pay for the license. It has a bit more features and has a native implementation backing the Java API.
There is always FOP (from apache) if you are willing to deal with XSLT and XSL-FO, but I believe they haven't updated those engines in a while.

I agree that iText is a great tool. However, the current version of iText is not free if you intend to use it in a closed source project. See Wikipedia:
In the end of 2009, iText version 5 is released under Affero GPL license. This license is drastically different from the previous license that iText had been distributed under, in that it requires anyone using iText 5 under a free license to provide the users with the full source of their application. Projects that do not want to provide their source code are required to purchase a commercial license for a non-disclosed price or they cannot upgrade to iText 5.
However, you may still use iText 4 under the LGPL license.

Take a look at Apache FOP. Very powerful.

IText will probably serve most of your purposes. However, if you are looking to convert from rtf or doc to pdf, you can use a java plugin for open source tools like OpenOffice( openoffice.org)
Hope this is helpful,
R

iText is probably your #1 standard in this area. You might also consider JODReports or Docmosis since they can do template-based reporting using standard word processor documents as templates.

Have you considered http://jasperforge.org/

Java Server question: Cheap/Mid-Level Creating report documents with designer (PDF/HTML)

What are some ways to create PDF reports in a Java server environment without having to use Java code to do so. Or maybe minimal Java code?
We have used iText and various htmltopdf solutions. Those work, but they take a lot of Java code create the documents and you have to code the positioning of all the elements?
Is there a solution that has a a designer tool? You design a report template with the designer and then deploy the template on a server?
We could pay for an enterprise solution.

You might be interested in JasperReports and iReport (which is sort of a designer IDE for JasperReports).

You can use JCopist to generate PDFs using FreeMarker templates rather than writing Java code. Another option that is mostly suitable to JSF-based projects, is to use JBoss Seam's iText-based template-based PDF generation tools: http://docs.jboss.org/seam/1.1.1.GA/reference/en/html/itext.html

You don't say if you're prepared to pay for an enterprise solution. If you are, then Thunderhead may be an option. It provides the means for templates (as you've specified) and can generate documents off the back of these in a variety of formats. You can interface to it via a variety of means (JMS / WebService / COM - not sure about the COM, on reflection). It has ActiveX controls to allow users to edit templates (with appropriate and fine-grained permissioning) and the template editing resembles a Word-based editor. It's very powerful indeed.

You should look at Docmosis. It uses standard word or openoffice documents as templates - so you design your document in a normal word processor. Depending on what you want to do in your templates this can be an ideal way to build reports since most developers (and users) already know how to work with word processors. You can then have Docmosis manipulate the document, merge in data and produce various formats. Have a look at the demo on the website.

any lib / api for filling in the field in microsoft doc?

I got a document that need to be filled in (it was in microsoft word doc), I have no idea how to filled in / integrated with my current web apps.
is there any good java api / lib that could be used ? preferrably the free one.
here is the example of doc that need to be filled in.
http://drop.io/callmeblessed/asset/debt-agremeent-certificate-doc

Apache POI - the Java API for Microsoft Documents

If Leniel's suggestion doesn't work (I would suggest trying POI first, as well), there's the OpenOffice.org java UNO API which has a different implementation. It introduces a significant runtime dependency, but if POI doesn't cut it, it's the obvious second choice.

Docmosis can do this as long as you have control of the source document and can specify placeholders etc. It's free and makes use of OpenOffice to do the format conversions. The Docmosis engine can do document manipulation (population, repetition, deletion etc). Load balancing and scalability features are paid for though.

Word document creation API in Java

I would like to create a word document using a template, replace some variables (fields) and save it as a new word document.
I was thinking using Apache POI, http://poi.apache.org/ is it the best for this purpose?
can you share your impression from it?

I've worked with POI before and it's certainly able to generate Word documents. But the devil is in the details.
Word has thousands of features: You can put numbered lists starting at #13 with negative indents into two joined cells of a table included in another table that is itself part of a bullet list... you get the idea. When the POI documentation says they are a work in progress, that reflects what will probably be an eternal state of trying to catch up to the (to us, undocumented) specification of Word.
Documents with a reasonably "normal" set of used features are well supported by POI, whose interfaces and methods are reasonable and consistent but sometimes require a bit of work. But as Pascal says, documents with a not too exorbitant set of features are also supported by RTF.
I have almost no experience "doing" RTF but it's probably a bit simpler than working with POI.
If you're working in an environment or for a customer who insists that your produced documents be .DOC rather than .RTF, then POI is pretty much your only choice, unless you can introduce a step where you use a bit of Office automation to convert RTF into DOC.
Update: I've had a couple more ideas in the meantime.
Using POI or creating RTF documents is something that you could do on practically any platform. At my job, all servers doing processing like this happen to be running Linux, for example.
However, in the likely case that your programs will run under Windows, there is another alternative: Jacob http://www.land-of-kain.de/docs/jacob/
Jacob is a COM interface for Java; it essentially allows you to "remote control" Windows programs such as Word and Excel. The document I linked to above is not to Jacob's own site but to a single page with "cookie cutter" recipes for using Jacob. The project itself is on SourceForge: http://sourceforge.net/projects/jacob-project/ But people claim, and rightly so, that the documentation is a bit lacking.
Jacob has the advantage over all other solutions that you're dealing with the "real" Word and therefore all capabilities of Word are available to you. This would be an alternative if there are detail aspects of your document that just can't be handled with POI or via the RTF format.

This is obviously way too late, But since 2013 there is a much better, more flexible solution to word document creation.
http://www.docx4java.org/trac/docx4j
I have had much more luck with docx4j than I ever did with POI.

I'm not sure of the exact status of the Word documents support in POI but, according to the POI website, work is still in progress (can't say what this mean exactly). So, at this time, I would not use POI but rather try to generate a RTF document. For this, you could :
Use RTFTemplate which is a RTF to RTF Engine that can generate RTF document as the result of the merge of a RTF model and data.
Use iText which is primarly a PDF generator but can also generate RTF.
Build your own custom solution (but I wouldn't do that).
I'd go for iText.

If you use a template, and do not want to create the word document from scratch, for what I know, POI is a pretty good solution. You open the template and select the zones you want to replace.
They say POI is still is developpement, but I've been using it in production environnement and it works pretty good at the moment.

I know this question is a bit old, but I think many people still find this with search engines, so I post another possibility to do what you want right here:
If the one and only goal is to have a Word Template and to replace some values in it, you might consider saving a Word Template as single xml (not docx) and then processing it with simple Java and without any Framework. If you want to do more (e.g. create lists or tables) you might also consider understanding the xml format and writing your own helpers before loading a Framework like POI.
Here is an example on how to do that:
http://dev-notes.com/code.php?q=10
This is the fast version, if you want a nice version, you could try using an XML processor.
PS: users might notice that the file extension is not doc but xml and they may blame you for that, but that's ok... just rename it to doc, word will recognize the format and everyone is happy again ;)

You should look into the Aspose.Words components. They have recently begun providing a Java version of the component.
See the following link: Aspose.Word for Java
This supports Word automation, creation and advanced features such as mail merging without the need for an instance of Microsoft Word on the machine. The real benefits are that you are able to work within the context of an actual word document and not having to compromise by creating RTFs etc.
The Java version is not currently as fully featured as the .Net version but the main core functionality is there and they are pushing very hard to have a feature equivalent version soon.
Also, if you purchase the Java version you get a years free upgrades / support as the new releases are created.

If you are working with docx documents, docx4j is an option. Like POI, its open source.

I created and use this: http://code.google.com/p/java2word