This software is a reference implementation of Etcetera Abduction. Given a knowledge base of first-order definite clauses and a set of observables, this software identifies the most probable set of assumptions that logically entails the observations, assuming the conditional independence of each assumption.

For more information, see the following:

• Gordon, Andrew S. (2016) Commonsense Interpretation of Triangle Behavior. In Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence (AAAI-16), February 12-17, 2016, Phoenix, Arizona.

This is a forked project from Andrew Gordon's EtcAbductionPy (https://github.com/asgordon/EtcAbductionPy).

The use of collections.deque makes S-expression parsing much faster.

For large knowledge base. This function requires python-mcdb (https://github.com/gstrauss/mcdb).

• --kbcache [FILE] or -kc [FILE]: create cache of knowledge base on FILE.
• --kbmcdb [FILE] or -km [FILE]: use cached knowledge base from FILE.

Example:

# Create a cache.
python etcabductionpy -k kb.lisp -kc kb.lisp.mcdb

# Use the cache for inference.
python etcabductionpy -km kb.lisp.mcdb ...

• --ilp or -l: use ILP solver (gurobi) to search the (n-)best solution(s).
• --ilp-show-nonetc or -lnonetc: output non-etc literals.
• --ilp-verbose or -lv: output an ILP solver log.