t3as - Term Vector Visualization

This directory contains 3 related projects:

bh_tsne: minor modifications to software from http://homepage.tudelft.nl/19j49/t-SNE.html
termvect: generates term vectors for all documents in a corpus (as input for bh_tsne)
clusteval: evaluates a cluster quality metric for labeled documents

Building

See the README for each project.

See the script termvect/src/main/scripts/run.sh for experiments using the 20NewsGroups data set.

See clusteval/3NewsGroups.pdf for results with 3 news-groups from the 20NewsGroups data set.

Visualization of document term-vectors using bh_tsne

Other

Language:C++ 66.8%Language:Scala 27.0%Language:Shell 4.2%Language:Python 2.1%