David Hope - D.Phil Candidate, University of Sussex Email

Research: Graphs applied to Natural Language Processing.

Supervisor: Dr. Bill Keller


Basic Graphs - in Flash


Sienna Graph
Weighted Edges Graph
Directed Weighted Edges Graph
Subsumption Graph
Clustering:

The max-max clustering algorithm (Keller & Hope, 2007) is used to cluster words.
A weighted clustering coefficient (Keller & Hope, 2009) is used to denote 'the importance' of each vertex within a graph.
Java WordNet::Similarity (beta)

A Java implementation of WordNet::Similarity - a Perl coded package that allows one to measure, in various ways, the similarity between word senses using the structure of WordNet.

Download Java WordNet::Similarity (beta.11.01)

Download Java WordNet::Similarity (beta.11.02) - thanks to Mihai Lintean @ University of Memphis for the bug fixes.

The Java version is simple to set up and use, though one should note it is in beta implementation; also, the Perl version has had quite a few years of input and testing and also comes with a lot of very thorough documentation so, use that if you can get it up and running: Perl WordNet::Similarity.


Google Web 1T (GW1T) Search Engine

A search engine for the GW1T. Allows a user to get counts for queries (inc. wildcard, proximity and Boolean queries).

One can also retrieve the entire contextual space for a term thus, it is feasible to construct semantic similarity measures for terms by, for example, using a vector space model.

Appraisal Theory Paper - Jonathon Read's research.
Language as song pattern: Dyson.

tc graphs


David Hope, University Of Sussex