Welcome to the SMILE-WIDE GitHub site

SMILE-WIDE is a Bayesian network library. Initially, SMILE-WIDE is a version of the SMILE library, augmented With Integrated Distributed Execution. This allows execution on very large datasets. As SMILE-WIDE is developed, BigData-specific capabilities will surpass the standard Bayesian network interfaces.

Using the library

Programmer-facing, SMILE-WIDE is a .jar library which you can include in your software. User-facing, it is also integrated into Hive as a UDF to provide posterior probabilities of missing values, given the observed values for each instance.

Languages used

SMILE-WIDE is written in Java, using the underlying SMILE library, which is written in C++. It uses Hadoop for inference on large data.

How to build the software

SMILE-WIDE is an Eclipse project configured to use Maven. All external dependencies are pulled from the appropriate Maven repositories. The code can be built from the IDE or directly from the command line.

Further information

For further information, including how to build, run, and test this software, see the READ-ME.html file included in the distribution (and also online in the GitHub Wiki for this project).