On the evaluation of the symbolic knowledge extracted from black boxes

As opaque decision systems are being increasingly adopted in almost any application field, issues about their lack of trans- parency and human readability are a concrete concern for end-users. Amongst existing proposals to associate human- interpretable knowledge with accurate predictions provided by opaque models, there are rule extraction techniques, ca- pable of extracting symbolic knowledge out of opaque mod- els. The quantitative assessment of the extracted knowledge’s quality is still an open issue. For this reason we provide here a first approach to measure the knowledge quality, encompass- ing several indicators and providing a compact score reflect- ing readability, completeness and predictive performance as- sociated with a symbolic knowledge representation. We also discuss the main criticalities behind our proposal, related to the readability assessment and evaluation, to push future re- search efforts towards a more robust score formulation.