cc.mallet.classify.tui.TODO Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of jcore-mallet-2.0.9 Show documentation

MALLET is a Java-based package for statistical natural language processing, document classification, clustering, topic modeling, information extraction, and other machine learning applications to text.

The newest version!

Make --argument=value work just as does --argument value
Fancier, nice spacing formating for usage printing
Feature selection for vectors2classify (no, FeatureSelectingClassifier will do)
Make vectors2info be able to print the instance list, like rainbow
All the complicated instace list splitting, as in rainbow
Specify the simple method without any 'new' or 'Trainer()'.
Specify all trainer meta-parameters via reflection, with syntax?
  --trainer MaxEnt,gaussianPriorVariance=10,numIterations=20
Implement student-T test for statistical significance in vectors2classify
Clean up all the logging statements so that training doesn't look so messy
Show progress with linefeed-less carriage returns, as in rainbow,
 unless the output is being piped to a file.  (How can you tell?)
Implement stemming for text2vectors: steal rainbow's implementation
Implement Vectors2Info, with all the things that rainbow does
Consider having a Vectors2Vectors that does splits, feature selection, etc.
 Perhaps all the instance list printing options actually go in here.
Implement Csv2Vectors and csv2vectors
Write web page documentation
Consider implementing single letter command-line options to go with long options.
Implement some fancier outputs, including graphs drawn by R.

Work with Nadia and/or Wei and/or Ron on a document classification task.


Use the same style indentation spacing as Emacs does!!!!  Tab only!!!

Bring in implementation of SVM
Implement KNN
Implement TFIDF classifier
Implement an ensemble method
Improve performance of Winnow
 (Compare this Winnow with MaxEnt, see if it wins, as did WhizBang's.)
Consider an architecture for active learning