Analogy Engines for Semantic Web

    We propose a new utility for Semantic Web called as Analogy Engine. Analogy engine employs an example based search approach to retrieve the most similar URIs for the given URI by comparing number of shared links. The Analogy engine is based on Analogy Space, which uses Singular Value Decomposition on matrix representation of a Semantic Network. However Analogy Space faces difficulty with networks having more than few thousand nodes. We present our preliminary work on scaling Analogy Space by dividing the network into multiple communities, and creating separate Analogy Space for each community. We show that this procedure results in significant improvements and can be used for a large scale network such as the Semantic Web.

Extended Divisi for creating Multiple Analogy Spaces:

We provide extended Divisi package which contains wrapper for CNM Algorithm, code to create multiple Analogy Spaces and code to work with RDF graphs.

Download extended Divisi

You will also need to download CNM algorithm (October 2008 Weighted version) compile it and place executables in  following directory in python folder
/lib/site-packages/csc/divisi

Following python packages are required:

• Networkx  
  
• Conceptnet  (If you wish to use Concept net Semantic Network)
  
• RDFLib   ( If you wish to handle Rdf data)

The Code has been tested on windows,  in order to use it on Linux you will need to
change the os.system calls by adding '\.' before the argument.

License:

The Code  provided is an extended version of the Divisi Packgage and is licensed under  GPL which is same as that of Divisi.
In case of comments, suggestion you can contact me on following email address.
Contact: Akshayubhat@gmail.com