Semantic Seriation based on Hamiltonian Path
Java
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
src/sg/edu/nus/comp/sseriation
test_data
.gitignore
AUTHORS
LICENSE
README

README

Semantic Seriation based on Hamiltonian Path is a package that takes a sparse term-document matrix provided in libsvm format, and calculates the seriation of the term space. Several distance functions are available. 

The JWI (the MIT Java Wordnet Interface) package is required to run seriation with WordNet-based distance functions. Make sure you adjust the WordNet directories in the source files.

Two files are typically required
1) Libsvm-formatted sparse term-document matrix with training instances.
2) Libsvm-formatted sparse term-document matrix with testing instances

The latter can be excluded, the seriation only relies on information from the first file. 

The classes under sg.edu.nus.comp.sseriation.order perform seriation with different distance functions. Drawing histograms of subsequent distances is also possible. The classes under sg.edu.nus.comp.sseriation.sdistance are helper functions to calculate WordNet-based distances.

If you use this code, please cite: 
Wittek, P., Darányi, S., Tan, C.L.: An Ordering of Terms Based on Semantic Relatedness. Proceedings of IWCS-09, 8th International Conference on Computational Semantics, pp. 235—247. Tilburg, The Netherlands. January, 2009.