Alchemy Lite is a software tool for representing and reasoning with probabilistic relational knowledge. This tutorial serves as a practical introduction to Alchemy Lite and assumes some familiarity with probability.
Alchemy Lite is based on Tractable Markov Logic (TML), a first-order probabilistic logic with guaranteed tractable inference. It allows one to build intuitive models in an object-oriented style while guaranteeing that inference will be efficient without resorting to approximation or ad hoc performance hacks. TML strikes a good balance between expressiveness and tractability, subsuming essentially all tractable models, including many high-treewidth ones.
The purpose of this tutorial is to introduce the reader to the key ideas of TML using standard examples, so that they can employ it in their own applications. The tutorial is not exhaustive, as at the time of writing this is a new line of research and there are still many possible applications that are unexplored, as well as several known examples that have not yet been implemented.