WordNet::Similarity::Visual - Perl extension for providing visualization tools


WordNet-Similarity-Visual documentation  | view source Contained in the WordNet-Similarity-Visual distribution.

Index


NAME

Top

WordNet::Similarity::Visual - Perl extension for providing visualization tools for WordNet::Similarity

SYNOPSIS

Top

Basic Usage Example

  use WordNet::Similarity::Visual;

  $gui = WordNet::Similarity::Visual->new;

  $gui->initialize;

DESCRIPTION

Top

This package provides a graphical extension for WordNet::Similarity. It provides a gui for WordNet::Similarity and visualization tools for the various edge counting measures like path, wup, lch and hso.

Methods

The following methods are defined in this package:

Public methods

$obj->new

The constructor for WordNet::Similarity::Visual objects.

Return value: the new blessed object

$obj->initialize

To initialize the Graphical User Interface and pass the control to it.

Discussion

The path measure defines the semantic similarity between two concepts as the inverse of length of the shortest path between the concepts in the hypernym trees of WordNet. This module displays the hypernym trees for both the concepts and the shortest path between these concepts.

The wup measure is based on the method proposed by Wu & Palmer and uses the depth of the two concepts in the hypernym tree and the depth of the Least Common Subscumer. It is based on the This module enables the user to view the hypertrees for the concepts. The lch measure implements a semantic measure proposed by Leacock & Chodrow. It uses the length of the shortest path between the two concepts and scales it by the maximum depth of the tree to compute the similarity score. For this measure this module displays the shortest path.

The hso measure measure computes the semantic relatedness between two concepts using the method proposed Hirst & St-Onge. They define the relatedness between two concepts based on the quality of links in the lexical chain connecting the two concepts.

The trace output from these measures is converted to a meta-language. This meta-language serves as the input ot the visualization module. The trace output is not used as the input to the visualization, because it might change in the furure versions of WordNet::Similarity, thus converting it to metalanguage prevents any of these changes to cause a major changes in the visualization module.

Meta-language

The first line in the meta language is the measure name. The next two line list all the possible shortest paths between the two concepts. The synsets represent the nodes along these paths, thile the relation names between these synsets represent the edges. If there is more than one shortest path they are also listed. The alternate shortest paths are seperated using the OR operator. The rest of the lines list all the other paths in the hypernym tree. These alternate hypernym trees also use the same system as used in the shortest path. The next line is the maximum depth of the hypertree

    path
    cat#n#1 hypernym feline#n#1 hypernym carnivore#n#1
    dog#n#1 hypernym canine#n#2 hypernym carnivore#n#1
    carnivore#n#1 hypernym placental#n#1 hypernym mammal#n#1 hypernym vertebrate#n#1 hypernym
      chordate#n#1 hypernym animal#n#1 hypernym organism#n#1 hypernym living_thing#n#1 hypernym
      object#n#1 hypernym entity#n#1 hypernym Root#n#1
    Max Depth = 13
    Path length = 5




SEE ALSO

Top

WordNet::Similarity WordNet::QueryData

Mailing List: <wn-similarity@yahoogroups.com>

AUTHOR

Top

Saiyam Kohli, University of Minnesota, Duluth kohli003@d.umn.edu

Ted Pedersen, University of Minnesota, Duluth tpederse@d.umn.edu

COPYRIGHT

Top


WordNet-Similarity-Visual documentation  | view source Contained in the WordNet-Similarity-Visual distribution.