Maven Setup

ClearTK is built with Apache Maven and we strongly recommend that you also build your ClearTK-based projects with Maven. See the Maven Getting Started Guide if you are not already familiar with Maven.

To use ClearTK in your Maven-based project, add something like the following to your pom.xml:

<properties>
  <cleartk.version>2.0.0</cleartk.version>
</properties>
...
<dependencies>
  <dependency>
    <groupId>org.cleartk</groupId>
    <artifactId>cleartk-ml</artifactId>
    <version>${cleartk.version}</version>
  </dependency>
  <dependency>
    <groupId>org.cleartk</groupId>
    <artifactId>cleartk-ml-liblinear</artifactId>
    <version>${cleartk.version}</version>
  </dependency>
  ...
</dependencies>

The example above declares dependencies on cleartk-ml, the main ClearTK machine learning interfaces, and cleartk-ml-liblinear, one of the more efficient implementations of these interfaces. For the full listing of ClearTK modules, see below.

ClearTK Modules

Machine learning

• cleartk-ml: the core machine learning APIs: classifiers, features, instances, feature extractors, feature encoders, etc.
• cleartk-ml-crfsuite: wrappers to use CRFsuite as a ClearTK classifier
• cleartk-ml-liblinear: wrappers to use LIBLINEAR as a ClearTK classifier
• cleartk-ml-libsvm: wrappers to use LIBSVM as a ClearTK classifier
• cleartk-ml-mallet: wrappers to use MALLET and GRMM as ClearTK classifiers
• cleartk-ml-opennlp-maxent: wrappers to use OpenNLP-maxent as a ClearTK classifier
• cleartk-ml-svmlight: wrappers to use SVM-light as a ClearTK classifier
• cleartk-ml-tksvmlight: wrappers to use Tree Kernels in SVM-light as a ClearTK classifier

Evaluation

• cleartk-eval: classes for evaluating pipelines - train/test, cross-validation, etc.

Type system

• cleartk-type-system: the official UIMA type system for ClearTK
• cleartk-corpus: readers and writers that operate on corpora using the ClearTK type system, including support for ACE 2005, CoNLL 2003, CoNLL 2005, Genia, PennTreebank, PropBank, TimeML, etc.
• cleartk-feature: feature extractors for the ClearTK type system, e.g. for extracting features from ClearTK tokens or ClearTK parse trees

Wrappers for external components

• cleartk-snowball: a wrapper around the Snowball stemmer
• cleartk-opennlp-tools: wrappers around the OpenNLP sentence segmenter, part-of-speech tagger, and syntactic parser
• cleartk-berkeleyparser: a wrapper around the Berkeley syntactic parser
• cleartk-clearnlp: a wrapper around ClearNLP, the successor to clearparser. Includes wrappers for its tokenizer, POS tagger, morphological analyzer (lemmatizer), dependency parser, and semantic role labeler.
• cleartk-maltparser: a wrapper around the Malt dependency parser
• cleartk-stanford-corenlp: a wrapper around the Stanford CoreNLP sentence segmenter, tokenizer, part-of-speech tagger, named-entity tagger, syntactic parser, dependency parser and coreference resolution system.

Home-grown components

• cleartk-token: sentence segmenters, tokenizers and part-of-speech taggers, including a segmenter based on java.text.BreakIterator, a PennTreebankTokenizer
• cleartk-timeml: models for extracting events, times and temporal relations, trained on TempEval 2013 data

Utility modules

• cleartk-test-util: test case base classes and utilities for testing licenses, parameter names, etc.
• cleartk-util: simple type-system agnostic readers and writers, and various utilities used by other ClearTK modules

Example code

Note: This module is only provided as source code. The code may change at any time. Never add cleartk-examples as a dependency.

• cleartk-examples: example code for part-of-speech tagging, BIO-chunking, bag-of-words document classification, etc.