Alchemy Lite
Software for Tractable Markov Logic
Alchemy Lite is a software package for inference in Tractable Markov Logic (TML),
the first tractable first-order probabilistic logic. It is an implementation of the
work in
A Tractable First-Order Probabilistic Logic by Pedro Domingos and Austin Webb.
(If you are not already familiar with TML, we recommend you read that paper first.)
Alchemy Lite allows for fast, exact inference for models formulated in terms of TML.
Alchemy Lite also has an optional interactive mode where the user can add new facts
to the knowledge base.
In the next release, we plan to include:
- learning algorithms for TML
- more complex inferences, such as parsing and entity resolution
- the addition of new TML rules in interactive mode
Alchemy Lite has been developed at the University of Washington by Chloe Kiddon and Pedro Domingos.
Alchemy Lite Tutorial
Download the code
Updates:
July 15, 2013 -- Initial version 1.0 is released.
Please send any questions, comments, or bugs to alchemy dash lite at cs dot washington dot edu.
Alchemy Lite is funded by: