CEL - A Polynomial-Time Reasoner for Life Science Ontologies

Abstract

CEL (Classifier for EL) is a reasoner for the small description logic EL which can be used to compute the subsumption hierarchy induced by EL ontologies. The most distinguishing feature of CEL is that, unlike all other modern DL reasoners, it is based on a polynomial-time subsumption algorithm, which allows it to process very large ontologies in reasonable time. In spite of its restricted expressive power, EL is well-suited for formulating life science ontologies. The Description Logic underlying CEL The system CEL is a first step towards realizing the dream of a description logic system that offers both sound and complete polynomial-time algorithms and expressive means that allow its use in real-world applications. It is based on recent theoretical advances that have shown that the description logic (DL) EL, which allows for conjunction and existential restrictions, and some of its extensions have a polynomial-time subsumption problem even in the presence of concept definitions and so-called general concept inclusions (GCI) [1]. The DL EL handled by CEL extends EL by so-called role inclusions (RI). On the practical side, it has turned out that the expressive power of EL is sufficient to express several large life science ontologies. In particular, the Systematized Nomenclature of Medicine (Snomed) [4] employs EL with RIs and acyclic concept definitions. The Gene Ontology (Go) [3] can also be expressed in EL with acyclic concept definitions and one transitive role (which is a special case of an RI). Finally, large parts of the Galen Medical Knowledge Base (Galen) [5] can be expressed in EL with GCIs and RIs. Because of the space limitations, we cannot introduce the syntax and semantics of EL in detail. We just mention the syntax elements, and illustrate their use by a small example. Full definitions can be found in [1, 2]. Like in other DLs, EL concepts are inductively defined starting with the sets of concept names NC and role names NR. Each concept name A is a concept, and so are the top concept >, conjunction C uD, and existential restriction ∃r.C. An EL ontology is a finite set of general concept inclusions (GCI) of the form C v D for concepts C,D, and complex role inclusions (RI) of the form r1 ◦ · · · ◦ rn v s for roles r1, . . . , rn, s. A primitive concept definition (PCDef) A v D is a GCI with the 1 CEL can be downloaded from http://lat.inf.tu-dresden.de/systems/cel/ . left-hand side a concept name, while a (non-primitive) concept definition (CDef) A ≡ D can be expressed using two GCIs. It is worthwhile to note that RIs generalize at least three expressive means important in bio-medical applications: role hierarchy, transitive role, and so-called right-identity axioms [4]. One of the most prominent inference problems for DL ontologies is classification: compute the subsumption hierarchy of all concept names occurring in the ontology. Endocardium v Tissue u ∃cont-in.HeartWall u ∃cont-in.HeartValve HeartWall v BodyWall u ∃part-of.Heart HeartValve v BodyValve u ∃part-of.Heart Endocarditis v Inflammation u ∃has-loc.Endocardium Inflammation v Disease u ∃acts-on.Tissue HeartDisease ≡ Disease u ∃has-loc.Heart

DOI: 10.1007/11814771_25

Extracted Key Phrases

1 Figure or Table

02040'04'05'06'07'08'09'10'11'12'13'14'15'16'17
Citations per Year

273 Citations

Semantic Scholar estimates that this publication has 273 citations based on the available data.

See our FAQ for additional information.

Cite this paper

@inproceedings{Baader2006CELA, title={CEL - A Polynomial-Time Reasoner for Life Science Ontologies}, author={Franz Baader and Carsten Lutz and Boontawee Suntisrivaraporn}, booktitle={IJCAR}, year={2006} }