Synthesizing Recursive Logic Programs by Inverting General Resolution

This repository contains the code and the dataset of benchmarking MPL and baseline.

build

1. prepare a UNIX environment, I use Alpine Linux
2. install Rust language
3. install clang compiler
4. clone this repository
5. run sh build.sh

experiments

1. completeness/soundness of mpl's abduction rule:
   • run ./v2/build/irctest.elf
2. deductive expressivity test with manually implemented program:
   • run ./v2/build/main.elf <taskPath>
   • <taskPath> are .ltr files located under ./v2/ex
3. performance benchmark:
   • run python3 -m dnc_ex dnc <taskName>

structure of the repository

• /dnc: source code of MPL inference engine (server)
• /dnc_kb: python controller (client)
• /dnc_ex: dataset generator and wrapper
• /v2: the new implementation of MPL engine