Project Overview

The S-Space Package is a collection of algorithms for building Semantic Spaces as well as a highly-scalable library for designing new distributional semantics algorithms. Distributional algorithms process text corpora and represent the semantic for words as high dimensional feature vectors. These approaches are known by many names, such as word spaces, semantic spaces, or distributed semantics and rest upon the Distributional Hypothesis: words that appear in similar contexts have similar meanings.

The research and development is being done by the Natural Language Processing group at UCLA led by David Jurgens and Keith Stevens, under the advisory of Dr. Michael Dyer.

See the Getting Started page for a quick introduction on how to use the S-Space package, see the Package Overview for information on the code and available features, or dive right into the Javadoc to see what's available now.

Goal

The goal for the S-Space Package is two fold:

to provide a uniform implementation for many common semantic space algorithms, and
to provide for a high-quality, scalable library for designing new distributional approaches.

We aim to provide a comprehensive framework for researchers to easily develop new algorithms without having to replicate much of the shared software. As a part of this, we have implemented a variety of utility libraries for interacting with matrices, vectors, parsers, clustering algorithms and SVD. Where possible, all libraries are implemented in Java for maximum portability. We also support a limited set of Java-bindings to native libraries primarily for high-performance SVD and clustering operations.

For those looking to implement their own Semantic Space algorithm within the S-Space package, we recommend looking at the Introduction page.

For more details see our ACL Systems Demonstration paper:

David Jurgens and Keith Stevens, (2010). The S-Space Package: An Open Source Package for Word Space Models. In System Papers of the Association of Computational Linguistics. Available here

Algorithms

For a list of the currently supported semantic space algorithms please see our algorithms page. Further resources for each algorithm may also be found on the publications page.

Other utilities include support for the Singular Value Decomposition, Clustering, and Hadoop-based algorithms.

Contact

For help using the S-Space Package, please contain our user mailing list: mailto:s-space-users@googlegroups.com. For question on development, bug reports or other code-specific questions, please contact our development mailing list: mailto:s-space-research-dev@googlegroups.com. If you have further questions, please feel free to contact David Jurgens or Keith Stevens.

The developers also frequently are on our IRC chat #airhead_research on irc.rizon.net

License and Restrictions

The S-Space software package is free software released under the GPL v. 2 license. See our license and restrictions page for full details.

Bookmark this on Delicious