eyeprolog

Explainable EV-depot configuration

Questions, not just filtering

A depot needs a charger compatible with its 60 kW available site power, 22 kW minimum charging rate, CCS2 connector, and three-phase supply.

EyeProlog does four useful things with the same rules:

  1. finds the compatible option (fleet22);
  2. recommends it;
  3. explains why each rejected option fails; and
  4. answers the reverse question: what would have to change to make this option viable?

Examples include increase_site_power_to(80), choose_charger_at_least_kw(22), and use_connector(ccs2).

That reversibility is a practical difference between a relational rule model and a one-way scoring/filtering pipeline. The rules describe the relationships that must hold, so they can support selection, diagnosis, and what-if questions without creating three separate business-rule implementations.

Files

Reproduce

rdf-to-prolog examples/data/explainable-ev-depot-configuration-input.nq \
  --rules examples/data/explainable-ev-depot-configuration-rules.pl \
  -o examples/explainable-ev-depot-configuration.pl

eyeprolog examples/explainable-ev-depot-configuration.pl

eyeprolog --goal write_results examples/explainable-ev-depot-configuration.pl \
  > examples/data/explainable-ev-depot-configuration-output.pl
prolog-to-rdf examples/data/explainable-ev-depot-configuration-output.pl \
  -o examples/data/explainable-ev-depot-configuration-output.nq
This site is open source. Improve this page.