Fitting Ontologies and Constraints to Relational Structures

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hosemann, Simon, Jung, Jean Christoph, Lutz, Carsten, Rudolph, Sebastian
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916905920495616
author Hosemann, Simon
Jung, Jean Christoph
Lutz, Carsten
Rudolph, Sebastian
author_facet Hosemann, Simon
Jung, Jean Christoph
Lutz, Carsten
Rudolph, Sebastian
contents We study the problem of fitting ontologies and constraints to positive and negative examples that take the form of a finite relational structure. As ontology and constraint languages, we consider the description logics $\mathcal{E\mkern-2mu L}$ and $\mathcal{E\mkern-2mu LI}$ as well as several classes of tuple-generating dependencies (TGDs): full, guarded, frontier-guarded, frontier-one, and unrestricted TGDs as well as inclusion dependencies. We pinpoint the exact computational complexity, design algorithms, and analyze the size of fitting ontologies and TGDs. We also investigate the related problem of constructing a finite basis of concept inclusions / TGDs for a given set of finite structures. While finite bases exist for $\mathcal{E\mkern-2mu L}$, $\mathcal{E\mkern-2mu LI}$, guarded TGDs, and inclusion dependencies, they in general do not exist for full, frontier-guarded and frontier-one TGDs.
format Preprint
id arxiv_https___arxiv_org_abs_2508_13176
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Fitting Ontologies and Constraints to Relational Structures
Hosemann, Simon
Jung, Jean Christoph
Lutz, Carsten
Rudolph, Sebastian
Artificial Intelligence
Databases
68T30 (Primary) 68P15, 03B70 (Secondary)
I.2.4; H.2.3
We study the problem of fitting ontologies and constraints to positive and negative examples that take the form of a finite relational structure. As ontology and constraint languages, we consider the description logics $\mathcal{E\mkern-2mu L}$ and $\mathcal{E\mkern-2mu LI}$ as well as several classes of tuple-generating dependencies (TGDs): full, guarded, frontier-guarded, frontier-one, and unrestricted TGDs as well as inclusion dependencies. We pinpoint the exact computational complexity, design algorithms, and analyze the size of fitting ontologies and TGDs. We also investigate the related problem of constructing a finite basis of concept inclusions / TGDs for a given set of finite structures. While finite bases exist for $\mathcal{E\mkern-2mu L}$, $\mathcal{E\mkern-2mu LI}$, guarded TGDs, and inclusion dependencies, they in general do not exist for full, frontier-guarded and frontier-one TGDs.
title Fitting Ontologies and Constraints to Relational Structures
topic Artificial Intelligence
Databases
68T30 (Primary) 68P15, 03B70 (Secondary)
I.2.4; H.2.3
url https://arxiv.org/abs/2508.13176