D-NCAS (http://www.ee.utsunomiya-u.ac.jp/~kawatalab/pse/ncas.html) is a distributed module-based problem solving.This distributed PSE is to support users to solve their problems on xomputers without complete knowledge for computing or without a full programming.

Cactus (http://www.cactuscode.org/)Cactus is an open source problem solving environment designed for scientists and engineers. Its modular structure easily enables parallel computation across different architectures and collaborative code development between different groups. Cactus originated in the academic research community, where it was developed and used over many years by a large international collaboration of physicists and computational scientists.

There are numerous other works relevant to machine learning and text that in various ways are listed as follows:

NLTK (http://nltk.sourceforge.net) also has "Classifier" and "ClassifierTrainer" classes for plug-and-play classifiers, as well as implementations of Naive Bayes, MaxEnt, feature selection, a "Token" class, finite state transducers with iterators over transitions. In addition it has facilities for tagging, parsing and information extraction.

OpenNLP (http://opennlp.sourceforge.net) is also a Java package intended for text processing. It also has rich Pipelines consisting of chains of individual pipe Foo2Bar component steps that can be arbitrarily configured and plugged together.

JavaNLP (http://www-nlp.stanford.edu/javanlp/) (from Chris Manning's group at Stanford.)

BioJava (http://www.biojava.org/). The BioJava Project is an open-source project dedicated to providing Java tools for processing biological data. This will include objects for manipulating sequences, file parsers, CORBA interoperability, DAS, access to ACeDB, dynamic programming, and simple statistical routines to name just a few things.

There is information about various Finite State Machine software at http://www.cs.jhu.edu/~jason/405/software.html, including pointers to AT&T Finite State package, which also has very general finite state transducers with iterators over transitions, arbitrary transition costs, generalized implementations of Viterbi and Forward Backward. In addition it has epsilon transitions, composition, and much more.

Weka: Plug-and-play machine learning components in Java http://www.cs.waikato.ac.nz/~ml/weka, including classes for "Classifier", "NaiveBayes", "DecisionStump", "LogisticRegression", etc. It also has methods for splitting training sets, and nice evaluation tools, and GUI components to boot.