OpenNLP / News: Recent posts

The forums are monitored and I make an effort to reply to everyone who posts. Unfortunatly, the monitoring services on sourceforge have not been very reliable about sending notices of forum activity. If you post and don't get a reply in a couple of days, please send a follow up email to my sourceforge account tsmorton@users.sf.net Thanks...Tom

This release improves the parser and coreference components, adds an n-gram package to improve training memory requirements for the parser, upgrades to the latest version of the maxent package, and adds preliminary support for Spanish. The upgrade to maxent-2.4.0 improves speed and reduces memory requirements for most components.
Enjoy!

This new release includes a coreference module as well as improvements to the existing sentence detector, tokenization, pos-tagging, and parsing modules. Enjoy!

One year, one month, and one day after the final release of the opennlp-common project we are releaseing the opennlp-tools package. This project consist of a combination of previous work release under the OpenNlp moniker as well as new work. The tools contain a sentence detector, a tokenizer, a pos-tagger, a chunker, a name finder, and a full parser.

Both opennlp.common and opennlp.hylo have been discontinued as packages in their own right. The final releases have been made only to allow the source code of their final state to be available in conjunction with release 0.7.0 of Grok. However, many things are living on in the OpenCCG project (http://openccg.sf.net), so go there for the latest.

At the moment, OpenNLP is thus only providing the organizational role for a number of different projects. See the website for details.

I've separated the hybrid logic stuff and created a new repository for it as a separate package. The CVS module is "Hylo".

Not much else for now. I'm looking at Gate, http://gate.ac.uk, to see if OpenNLP and Gate can join forces on the points at which we overlap. They have some really nice stuff!

So at this point it looks like the API has made it through the rather wide-sweeping changes that have been going on since October, so I have bumped the version from 0.6.6 to 0.8.0 to reflect that.

Of course, the develoment will continue to move on!

The XML representation for information about a text being processed through a pipeline (via an NLPDocument object) has changed considerably.

The interfaces will continue to go through much more cleaning and simplification over the coming month.

0.6.4 has some nice new aspects which can be checked out in the notes for the release. In particular, unification is being handled differently than before and XmlUtils is back in this release and working with JDOM. Expect lots of changes in the opennlp.common.unify and opennlp.common.xml packages over the next few weeks.

So, I finally got around to restructuring the build system of OpenNLP (and various other projects) and making a proper release. In addition to the jar file, there is also a tar gzipped file containing all the source and the supporting libraries for OpenNLP.

There are a number of changes from version 0.4.0, most of which have been available in the CVS for a long time. While thsee are substantial, the OpenNLP API is still nowhere near what it should be. However, expect improvements in the unification stuff and in the way that XML is used in the preprocessing architecture. With the creation of the Leo Project (leonlp.sf.net), this API has the potential to become much more general and useful.... read more

This is just a quick note to say that though I have failed to make a release, there are changes in the OpenNLP interfaces in the CVS. The primary difference is that the interfaces have been split into several subpackages such as quipu.opennlp.discourse, quipu.opennlp.parse, etc. Another difference is that I am trying to move the interfaces away from being categorial grammar specific.

There another interesting thing to mention on the future of OpenNLP. I am interested in moving much of the focus of OpenNLP to be the definition of data formats (generally XML) for resources such as lexicons and document markup by language processors. I'll be working for and with NASA this summer on defining XML specifications for lexicons, and hopefully this will become part of the OpenNLP project. The major contribution of this emphasis is to move away from the Java-centric nature of OpenNLP as it currently is.... read more