Symbolic AI for XAI - Evaluating LFIT Inductive Programming for Explaining Biases in Machine Learning

Aug 2, 2026

Uploaded by Noah

Machine learning methods increasingly support biometrics and personal information processing across forensics, e-health, recruitment, and e-learning, where white-box, human-readable explanations are essential. The work studies inductive logic programming and the LFIT technique, aiming to derive a declarative, propositional-logic theory equivalent to a black-box model under given conditions. A methodology is evaluated for fair recruitment using gender and ethnicity in automated CV ranking, and expressiveness is preliminarily tested on adult income data to assess how higher incomes may be biased by demographics.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages26

Reading Time1 h 6 min

AccessFree to download

Introduction

Motivation for Explainable AI

Explanations as a Requirement in Key Domains

Study Focus: Recruitment and Income Bias

Abstract

LFIT and Inductive Logic Programming

Fairness Use Case: CV Ranking with Demographics

Preliminary Expressiveness Test on Adult Income Data

computers Article Symbolic AI for XAI: Evaluating LFIT Inductive Programming for Explaining Biases in Machine Learning Alfonso Ortega 1,2, *, Julian Fierrez 2, Aythami Morales 2, Zilong Wang 3, Marina de la Cruz 4, César Luis Alonso 1 and Tony Ribeiro 5,6 􀀁􀀂􀀃􀀁􀀄 􀀆􀀇􀀈 􀀁􀀂􀀃􀀄􀀅􀀆􀀇 Citation: Ortega, A.; Fierrez, J.; Morales, A.; Wang, Z.; de la Cruz, M.; Alonso, C.L.; Ribeiro, T. Symbolic AI for XAI: Evaluating LFIT Inductive Programming for Explaining Biases in Machine Learning. Computers 2021, 10, 154. [https://doi.org/10.3390/](https://doi.org/10.3390/)[ ](https://doi.org/10.3390/)computers10110154 Academic Editor: Ana Filipa Sequeira Received: 11 September 2021 Accepted: 8 November 2021 Published: 17 November 2021 Publisher's Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ([https://](https://)[ ](https://)[creativecommons.org/licenses/by/](creativecommons.org/licenses/by/)[ ](creativecommons.org/licenses/by/)[4.0/](4.0/)) . 1 Departamento de Informática, Universidad de Oviedo, 33204 Oviedo, Spain; [calonso@uniovi.es](calonso@uniovi.es) 2 Escuela Politécnica Superior, Universidad Autónoma de Madrid, 28049 Madrid, Spain; julian.ﬁ[errez@uam.es](errez@uam.es) (J.F.); aythami.morales@uam.es (A.M.) 3 Institute IMDEA Software, 28223 Pozuelo de Alarcón, Spain; [zilong.wang@imdea.org](zilong.wang@imdea.org) 4 Escuela Superior de Ingeniería y Tecnología (ESIT), Universidad Internacional de la Rioja, 26006 Logroño, Spain; [marina.delacruz@unir.net](marina.delacruz@unir.net) 5 Laboratoire des Sciences du Numérique de Nantes, 44300 Nantes, France; [tony.ribeiro@ls2n.fr](tony.ribeiro@ls2n.fr) [6](6 National Institute of Informatics)[ National Institute of Informatics](6 National Institute of Informatics), [Tokyo 101-8430](Tokyo 101-8430), [Japan](Japan) * Correspondence: alfonso.ortega@uam.es Abstract: Machine learning methods are growing in relevance for biometrics and personal information processing in domains such as forensics, e-health, recruitment, and e-learning. In these domains, white-box (human-readable) explanations of systems built on machine learning methods become crucial. Inductive logic programming (ILP) is a subﬁeld of symbolic AI aimed to automatically learn declarative theories about the processing of data. Learning from interpretation transition (LFIT) is an ILP technique that can learn a propositional logic theory equivalent to a given black-box system (under certain conditions) . The present work takes a ﬁrst step to a general methodology to incorporate accurate declarative explanations to classic machine learning by checking the viability of LFIT in a speciﬁc AI application scenario: fair recruitment based on an automatic tool generated with machine learning methods for ranking Curricula Vitae that incorporates soft biometric information (gender and ethnicity) . We show the expressiveness of LFIT for this speciﬁc problem and propose a scheme that can be applicable to other domains. In order to check the ability to cope with other domains no matter the machine learning paradigm used, we have done a preliminary test of the expressiveness of LFIT, feeding it with a real dataset about adult incomes taken from the US census, in which we consider the income level as a function of the rest of attributes to verify if LFIT can provide logical theory to support and explain to what extent higher incomes are biased by gender and ethnicity. Keywords: explainable artiﬁcial intelligence; inductive logic programming; fair recruitment; fair income level; propositional logic 1. Introduction Statistical and optimisation-based machine learning algorithms are supported by well-known and solid numerical and statistical methods. These t

FAQs

What does the paper evaluate regarding LFIT and explainable AI?

It evaluates LFIT inductive programming as a way to incorporate accurate, human-readable declarative explanations derived from machine learning systems, under stated conditions.

How is fairness addressed in the main application scenario?

Fairness is examined through an automatic recruitment tool that ranks curricula vitae using soft biometric information such as gender and ethnicity.

What preliminary test is conducted to study LFIT expressiveness beyond recruitment?

The authors run LFIT on a real adult income dataset from the US census, treating income level as a function of other attributes to investigate potential bias by gender and ethnicity.

Aug 2, 2026

Uploaded by Noah