Skip to content

Latest commit

 

History

History
104 lines (73 loc) · 4.12 KB

README.md

File metadata and controls

104 lines (73 loc) · 4.12 KB
Raw

C-Cat Wordnet - A simple Java library for Princeton Wordnet

Introduction

C-Cat Wordnet is a java framework for using and editing the Princeton Wordnet lexical ontology. It includes a simple to use tools for retrieving synsets, comparing synsets with similarity measurse, disambiguating sentences, and directly modifying the synset hierachy to include new information.

Build and Installation

C-Cat Wordnet uses maven for compilation and dependency management. It depends a sister sub-module for the larger C-Cat package (util). After installing the C-Cat util package, run the following from the wordnet directory:

mvn package

This will create an all inclusive jar that includes the C-Cat Wordnet library, all depenendencies, and a copy of the Princeton Wordnet 3.0 dictionary files.

Using Wordnet

Key Components

The Wordnet library centers around two core concepts: a OntologyReader and a set of Synsets. The OntologyReader reads a Wordnet database from files on disk or included in the running classpath and creates an in memory structure mapping lexical units to Synsets. Each Synset connects together words that have a shared meaning and is further connected to other Synsets that are semantically related. For instance, one Synset may collect 'child', 'kid', and 'youngster' together and link to the more abstract Synset for 'juvenile'.

The default OntologyReader reads the standard Princeton Wordnet dictionary files and is named WordNetCorpusReader. You can load the database using two simple methods:

OntologyReader reader = WordNetCorpusReader.initialize("/path/to/dict");

or

OntologyReader reader = WordNetCorpusReader.initialize("dict", true);

The first method simply reads the dictionary files from the specified path and loads the entire hierarhcy into memory. The second method makes use of the dictionary files included in the all inclusive jar and reads the dictionary files from within the jar itself. This shows the simplest example for loading the database from within nearly any environment.

Making Simple Queries

The OntologyReader provides several methods for accessing Synsets given words. To get all possible Synsets for the word 'cat', you can do the following:

Synset[] allSynsets = reader.getSynsets("cat");
Synset[] nounSynsets = reader.getSynsets("cat", PartsOfSpeech.NOUN);
Synset cat1 = reader.getSynset("cat", PartsOfSpeech.NOUN, 1);

Each example shows how to access increasingly more specific Synsets for a word. The first returns any Synset associated with 'cat'. The second returns only noun Synsets associated with 'cat'. The last returns only the first noun Synset for 'cat'.

Once you've obtained a synset, you can access a variety of details:

// Get the definition
cat1.getGloss();

// Iterate through the lemmas
for (Lemma lemma : cat1.getLemmas())
  System.out.println(lemma);

// Print the Synset name.  This is a simple reference for the Synset.
System.out.println(cat1.getName());

// Print the name of each parent Synset
for (Synset parent : cat1.getParents())
  System.out.println(parent.getName());

// Print the name of any related Synset
for (Synset related : cat1.allRelations())
  System.out.println(related.getName());

You can furthremore modify the synset with new details:

// Change the definition
cat1.setDefinition("A cute fluffy pet");

// Add a new related Synset into the hierachy.
Synset newSyn = new BaseSynset("nori.n.1");
cat1.addRelation(Relation.HYPONYM, newSyn);
newSyn.addRelation(Relation.HYPERNYM, cat1);

Saving Your Modifications

The C-Cat Wordnet package makes it easy to save these changes to any format you can design. If you want to save the changes we've made to the database above, you can simply do:

OntologyWriter writer = new WordNetCorpusWriter();
writer.saveOntology(reader, "/path/to/new/dict");

This will save the modified hierarchy to the specified path in the standard Princeton Wordnet database format, which can easily be loaded again using the C-Cat Wordnet package or nearly any other library for the Princeton Wordnet.