Descriptive Logic for the Semantic Web Introduction

Descriptive Logic for the Semantic Web

Introduction • The vision of a Semantic Web has recently drawn considerable attention, both from academia and industry. • Description Logics are often named as one of the tools that can support the Semantic Web and thus help to make this vision reality.

Introduction • Descriptions Logics are very useful for defining, integrating, and maintaining ontologies, which provide the Semantic Web with a common understanding of the basic semantic concepts used to annotate Web pages.

The Semantic Web and Ontologies • World Wide Web has become an indispensable means of providing and searching for information. • Searching the Web in its current form is, however, often an infuriating experience since today's search engines usually provide a huge number of answers, many of which are completely irrelevant, whereas some of the more interesting answers are not found.

The Semantic Web and Ontologies • One of the reasons for this unsatisfactory state of affairs is that existing Web resources are usually only human understandable: the mark-up (HTML) only provides rendering information for textual and graphical information intended for human consumption.

The Semantic Web and Ontologies • The Semantic Web aims for machine -understandable Web resources, whose information can then be shared and processed both by automated tools, such as search engines, and by human users.

The Semantic Web and Ontologies • This sharing of information between different agents requires semantic mark-up, i. e. , an annotation of the Web page with information on its content that is understood by the agents searching the Web.

The Semantic Web and Ontologies • To make sure that different agents have a common understanding of this semantic mark-up, one needs ontologies that establish a joint terminology between the agents.

The Semantic Web and Ontologies • Basically, an ontology is a collection of definitions of concepts and the shared understanding comes from the fact that all the agents interpret the concepts w. r. t. the same ontology.

Ontology Language • The use of ontologies in this context requires a well-designed, well-defined, and Web-compatible ontology language with supporting reasoning tools. • The syntax of this language should be both intuitive to human users and compatible with existing Web standards (such as XML, RDF and RDFS). • Its semantics should be formally specified since otherwise it could not provide a shared understanding. • Finally, its expressive power should be adequate,

Employing Reasoning • Reasoning is important to ensure the quality of an ontology. It can be employed in different development phases: 1. During ontology design, it can be used to test whether concepts are non-contradictory and to derive implied relations.

Employing Reasoning 2. One usually wants to compute the concept hierarchy since information on which concept is a specialization of another and which concepts are synonyms is very useful when searching Web pages annotated with such concepts.

Employing Reasoning 3. Since it is not reasonable to assume that there will be a single ontology for the whole Web, interoperability and integration of different ontologies is also an important issue. Integration can, for example, be supported by asserting inter-ontology relationships and testing for consistency and computing the integrated concept hierarchy.

Employing Reasoning 4. reasoning may also be used when the ontology is deployed, i. e. , with Web pages that are already annotated with its concepts. • One can, for example, determine the consistency of facts stated in the annotation with the ontology or infer instance relationships.

Description Logics • Description logics (DLs) are a family of knowledge representation languages that can be used to represent the knowledge of an application domain in a structured and formally well-understood way.

Why “Description” Logics • The important notions of the domain are described by concept descriptions, i. e. , expressions that are built from atomic concepts (unary predicates) and atomic roles (binary predicates) using the concept and role constructors provided by the particular DL. • On the other hand, DLs differ from their predecessors, such as semantic networks and frames, in that they are equipped with a formal, logic-based semantics.

Example • Assume that we want to define the concept of “A man that is married to a doctor and has at least five children, all of whom are professors. ” Human ¬Female married. Doctor (≥ 5 child) • Boolean constructors conjunction ( ), which is interpreted as set intersection. • Negation (¬), which is interpreted as set complement • Existential restriction constructor ( R: C) • Value restriction constructor ( R: C) • Number restriction constructor (≥ n R) child

Example • An individual, say Bob, belongs to married. Doctor iff there exists an individual that is married to Bob (i. e. , is related to Bob via the married role) and is a doctor (i. e. , belongs to the concept Doctor). • Bob belongs to (≥ 5 child) iff he has at least five children, and he belongs to child: Professor iff all his children (i. e. , all individuals related to Bob via the child role) are professors.

Terminological Axioms • In addition to this description formalism, DLs are usually equipped with a terminological and an assertional formalism. • In its simplest form, terminological axioms can be used to introduce names (abbreviations) for complex descriptions. • For example, we could introduce the abbreviation Happy. Man for the concept description from above. • child. Human says that only humans can have human children.

Assertional Formalism • The assertional formalism can be used to state properties of individuals. For example, the assertions Happy. Man(BOB), child(BOB; MARY) state that Bob belongs to the concept Happy. Man and that Mary is one of his children.

Subsumption Description Logic systems provide their users with various inference capabilities that deduce implicit knowledge from the explicitly represented knowledge. • The subsumption algorithm determines subconcept-superconcept relationships: C is subsumed by D iff all instances of C are necessarily instances of D, i. e. , the first description is always interpreted as a subset of the second description.

Subsumption • For example: Happy. Man is subsumed by child. Professor since instances of Happy. Man have at least five children, all of whom are professors, they also have a child that is a professor.

Instance • The instance algorithm determines instance relationships: the individual i is an instance of the concept description C iff i is always interpreted as an element of C. • For example, given the assertions from above and the definition of Happy. Man, MARY is an instance of Professor.

Consistency • The consistency algorithm determines whether a knowledge base (consisting of a set of assertions and a set of terminological axioms) is non-contradictory. • For example, if we add ¬ Professor(MARY) to the two assertions from above, then the knowledge base containing these assertions together with the definition of Happy. Man from above is inconsistent.

Expressivities vs. Complexities • In order to ensure a reasonable and predictable behavior of a DL system, these inference problems should at least be decidable for the DL employed by the system, and preferably of low complexity. • Consequently, Investigating this trade-off of the expressive power DL in question mustexpressivities ofin anand between the be restricted DLs appropriate the complexity of their inference way. problems has been one of the most important issues in DL research.

Expressivities vs. Complexities Roughly, the research related to this issue can be classified into the following four phases 1. Phase 1 (1980 -1990) was mainly concerned with implementation of systems, such as KLONE, K-REP, BACK, and LOOM. These algorithms are usually very efficient (polynomial), but they have the disadvantage that they are complete only for very inexpressive DLs, i. e. , for more expressive DLs they cannot detect all the existing subsumption/instance relationships.

Expressivities vs. Complexities 2. Phase 2 (1990 -1995) started with the introduction of a new algorithmic paradigm into DLs, so-called tableaubased algorithms. 3. Phase 3 (1995 -2000) is characterized by the development of inference procedures for very expressive DLs, either based on the tableau-approach or on a translation into modal logics.

Expressivities vs. Complexities 4. We are now at the beginning of Phase 4, where industrial strength DL systems employing very expressive DLs and tableau-based algorithms are being developed, with applications like the Semantic Web or knowledge representation and integration in bio-informatics in mind.

Description Logics as Ontology Languages High quality ontologies are crucial for the Semantic Web, and their construction, integration, and evolution greatly depends on the availability of a well-dened semantics and powerful reasoning tools. • For an ontology language for the Semantic Web, a W 3 C ontology standard, called DAML+OIL, has a syntax based on RDF Schema. • It can be translated into expressive DL SHIQ and the developers have tried to find a good compromise between expressiveness and the complexity of reasoning.

Description Logics as Ontology Languages • Although reasoning in SHIQ is decidable, it has a rather high worstcase complexity (Exp. Time). • Nevertheless, there is a highly optimized SHIQ reasoner (Fa. CT) available, which behaves quite well in practice.

SHIQ makes reasoning hard 1. SHIQ provides number restrictions that are far more expressive than before: 1. qualified number restrictions -- being able to say that a person has at most two children (without mentioning the properties of these children): (≤ 2 child), one can also specify that there is at most one son and at most one daughter: (≤ 1 child. ¬ Female) (≤ 1 child. Female)

SHIQ makes reasoning hard 2. SHIQ allows the formulation of complex terminological axioms like “humans have human parents”: Human parent. Human • • To test consistency of the assertion Human(BOB) under this constraint, a naive tableau-based algorithm would not terminate since it would generate an innite chain of ancestors of Bob. To obtain a terminating procedure, one must check the computation for cycles by blocking the generation of role successors under certain conditions.

SHIQ makes reasoning hard Tableau methodology* • • Invented in the 1950's by Beth and Hintikka and later perfected by Smullyan and Fitting, is today one of the most popular proof theoretical methodologies. Firstly because it is a very intuitive tool, and secondly because it appears to bring together the proof-theoretical and the semantical approaches to the presentation of a logical system. * The Handbook of Tableau Methods. Edited by Marcello D'Agostino, Dov M. Gabbay, Reiner Hähnle, and Joachim Posegga. Kluwer Academic Publishers, 1999.

SHIQ makes reasoning hard Tableau methodology • • Recent advances in tableau-based theorem proving have drawn attention to tableaux as a powerful deduction method for classical first-order logic, in particular for non-clausal formulas accommodating equality. There is a growing need for a diversity of nonclassical logics which can serve various applications, and for algorithmic presentations of these logicas in a unifying framework which can support (or suggest) a meaningful semantic interpretation.

SHIQ makes reasoning hard Example: Sentence Tableaux* • Sentence tableaux provide us with a way of telling whether a set of sentences is consistent or not. They work in effect by breaking down sets of complex sentences into sets of simpler sentences where inconsistency is easier to spot. * Sentence Tableaux

SHIQ makes reasoning hard Example: Sentence Tableaux • Let us investigate the consistency of the following set of sentences: John likes rugby if he is Welsh John is Welsh unless his father is Scottish John's father is Welsh but John does not like rugby.

SHIQ makes reasoning hard • • Example: Sentence Tableaux Step 1: Translate the English sentences using the truth-functors of the propositional calculus. (Remember to use stand-alone sentences for the constituents. ) This gives us: [John is Welsh → John likes rugby] [John is Welsh ∨ John's father is Scottish] [John's father is Welsh ∧ ¬John likes rugby]

SHIQ makes reasoning hard • • • Example: Sentence Tableaux Step 2: Abbreviate the short constituent English sentences with single capital letters. W: John is Welsh Called “interpretation” L: John likes rugby F: John's father is Welsh S: John's father is Scottish “Sentence Letter”

SHIQ makes reasoning hard Example: Sentence Tableaux • Become: [W → L] [W ∨ S] [F ∧ ¬L] • In each case the resulting combinations of letters and truth-functors is a formula, (sometimes called a "wff": i. e. well formed formula; pronounced "woof". ); the process of translating into formulae is formalisation.

SHIQ makes reasoning hard Example: Sentence Tableaux • Step 3: Construct the tableau. To do this: (a) List the original formulae. (b) Develop the tableau according to the tableau rules.

SHIQ makes reasoning hard • Example: Sentence Tableaux The tableau takes the form of an upside down tree, with the original formulae constituting the trunk. The rules tell us – how to extend branches – how to close branches. • Each path from root to tip is called a "branch". So each branch actually includes the trunk.

SHIQ makes reasoning hard • • • Example: Sentence Tableaux Each such branch, until it is closed, represents a way in which the original sentences might perhaps be true. Closing a branch signifies that this branch is not after all a way in which the original sentences could all be true. So if all the branches close, that means that there was no way in which the original sentences could all be true; that is, that they are inconsistent.

SHIQ makes reasoning hard • • Example: Sentence Tableaux There is just one rule for closing a branch: we may close a branch just if it contains both a formula and the same formula with a "¬" in front. That is to say, a branch closes just if it contains both " φ " and "¬ φ " for some φ. To close a branch we just underline the tip.

SHIQ makes reasoning hard • • Example: Sentence Tableaux Let us now construct a tableau for our example. We apply the rule like this: [W → L] √[W ∨ S] [F ∧ ¬L] W S

SHIQ makes reasoning hard • Example: Sentence Tableaux After each application of a rule for extending branches, we check whether we can close a branch. [W → L] √[W ∨ S] √[F ∧ ¬L] W S F ¬L There are two ways in which the original sentences might all be true: if "[W →L]", "W", "F" and "¬L" are all true, and if "[W →L]", "S", "F" and "¬L" are all true.

SHIQ makes reasoning hard • Example: Sentence Tableaux Last step: √[W → L] √[W ∨ S] √[F ∧ ¬L] W S F ¬L ¬W L

SHIQ makes reasoning hard • Example: Sentence Tableaux Close branch: √[W → L] √[W ∨ S] √[F ∧ ¬L] W S F ¬L ¬W L

SHIQ makes reasoning hard • • • Example: Sentence Tableaux Closing a branch tell us that this branch does not after all represent a way in which the original sentences could all be true; because there is no way for "φ" and "¬φ" both to be true. We now have just one open branch. So the tableau now tells us that there is just one way in which the original sentences could all be true. They would be true if the following were all true: "S", "F", "¬L" and "¬W".

SHIQ makes reasoning hard • • Example: Sentence Tableaux If the branches were all closed, we could, of course, conclude that the original sentences could not all be true. But the fact that we have an open branch and the tableau has finished does not tell us that the sentences could all be true.

SHIQ makes reasoning hard • • Example: Sentence Tableaux We must go back to the interpretation to see if the unticked formulae in our open branch could in fact all be true. Need to decide whether the following could be true together: "John's father is Scottish", "John's father is Welsh", "John does not like rugby" and "John is not Welsh".

SHIQ makes reasoning hard • • Example: Sentence Tableaux Perhaps we will reckon that it is impossible to be both Welsh and Scottish. In which case we will say that these cannot be true together; so nor can the original sentences; so the original set is inconsistent. Perhaps we will reckon that if one's father is Welsh, one must be Welsh as well. In which case again we will conclude that the original set is inconsistent. Otherwise, presumably, we will conclude that the original set is consistent.

SHIQ makes reasoning hard 3. SHIQ also allows for inverse roles; for example, in addition to child one can also use its inverse parent, with the connection between a role and its inverse being taken into account during reasoning.

End