Merge, Explain, Iterate: A Combination of MHS and MXP in an ABox Abduction Solver

被引：0

作者：

Homola, Martin ^{[1
]}

Pukancova, Julia ^{[1
]}

Boborova, Janka ^{[1
]}

Balintova, Iveta ^{[1
]}

机构：

[1] Comenius Univ, Bratislava 84248, Slovakia

来源：

LOGICS IN ARTIFICIAL INTELLIGENCE, JELIA 2023 | 2023年 / 14281卷

关键词：

Abduction; Description logics; Ontologies; OWL;

D O I：

10.1007/978-3-031-43619-2_24

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Minimal Hitting Set (MHS) is a well-known and complete method to compute all minimal explanations of an ABox abduction problem in Description Logics (DL). MHS is NP-complete and generally recognized as inefficient. We leverage on MergeXplain (MXP) which is fast but incomplete - by combining it with MHS in a hybrid algorithm MHS-MXP to regain completeness. In this paper, we describe: (a) the underlying theory to establish the completeness of MHS-MXP and show its relevant properties; (b) a class of inputs on which MHS-MXP has the greatest advantage; (c) an experimental implementation; (d) an empirical evaluation on both favourable and unfavourable inputs.

引用

页码：338 / 352

页数：15