Report on Ontology spectrum of Knowledge Representation & Semantic Web Technologies

In the same way, McGuinness (Lassila & McGuinness, 2001) has categorized the ontologies based on their expressiveness, grounded on the information mentioned that there is a requirement for the expression of ontology. Therefore, ontology can be seen based on the interpretation types related to the word as described in figure 3. We could distinguish between a less or further complicated ontologies' concepts which could be ranged from a controlled vocabulary, a glossary, to reach the entire ontologies that offer broad logical limitations such as part, inverse, disjoints, etc. Below are the points that distinguished within the spectrum:

·         Controlled vocabularies: The simplest potential ontology’s concept described as vocabulary, which defines as a restricted list of terms

·         Glossary: This is a list of terms along with the meanings, whereas the meanings would typically communicate within natural language speech and subject to the human. This type of speech is unable to be used by computer devices since they are quite vague.

·         Thesauri: Thesauri do not provide any clear hierarchical system, even though this could frequently be assumed by narrower or broader specifications of the term. In a thesaurus, the definitions of meanings could be construed by a computer device.

·         Informal Is-a hierarchies: These are the ontologies in which a broad concept of both specialization and generalization is provided while it does not acquire a stringent subclass hierarchy. Yahoo! Is one of the classic examples which provides high-level classifications at a slight number, but does not acquire a clear hierarchical system.

·         Formal Is-a hierarchies: The ontologies here are organized in rendering to a stringent subclass hierarchy. Thus, an inheritance for these ontologies is often applied regarding the vase that if a C concept is a superclass of the C’ concept, then any classification of C should automatically be a subclass of C’.

·         Formal instances: The ontologies that contain formal instance connections are considered as the ontologies’ natural extension which applying a stringent hierarchical system. Once there are formal instance connections, the ontologies will also comprise the content from ground individuals along with their relationship with the concepts that they instantiate.

·         Frames (concept properties’ description): These are the ontologies that defined by their properties’ features. For instance, the types of features like author, title, and also the publisher may define a Book concept. The properties included within the concept description turn out to be further interesting once these properties could apply inheritance. Therefore, a further general concept can be specified for these properties and be innate down the hierarchy by further particular concepts.

·         Value restriction: There are restrictions applied in these ontologies on the values associated with the properties. For instance, we could limit the quantity related to the property of the Author to be poised of two names utmost in defining the concept Book. These limitations are typically inherited by the sub-concepts of the concept in which they are declared since the first time. A problem might be posed when the ontology which is not stringent subclass relation supports the kind of hierarchical connection.

·         General logical constraints: A quite expensive languages present these ontologies such as Ontolingua (Farquhar et al. 1997) that let the specification of the initial order logic constraints on the concepts along with their properties. For instance, the properties may be articulated as logical speech or could be grounded on the mathematical equations which utilize the other properties’ values.

3.5.3    Web Ontology Language (OWL)

OWL is a logic-based knowledge representation language intended for describing and instantiating ontologies widely used in the semantic web. OWL was standardized byW3C in 2004, and subsequently, in 2012 its most recent revision, OWL 2, became a W3C recommendation which started by the World Wide Web Consortium (W3C, 2012). It was shaped on the top of RDF, whereas the RDF Schema to comprise their restrictions and the vocabulary extensions meant for cardinality constraints features of richer property, etc. OWL could also demonstrate the complicated semantics as an extended vocabulary of both RDF and RDFS applying the properties, classes, cardinality, and relationships. It is designed for the applications that have to do the information content process, symbolize complicated and rich knowledge around the things, the group of things, along with the relationship among them, rather than only presenting some information to the human and accommodates better machine interpretability of the Web content. OWL ontology engineers consequently can employ an automated deduction structure named with 'reasoners,’ with a purpose to infer the innovative knowledge from their ontologies. Further clearly, a ‘reasoner’ creates simplicity asserted to clear the knowledge, thus, letting the engineers verify (some forms of) results of their declared axioms. 

Three areas influenced the OWL design such as , , as well as the  (Horrocks, Patel-Schneider, McGuinness, & Welty, 2010).

·         Description Logics: It has a purpose to convey the reasoning and expressive energy to the Semantic Web. Description logics are a relative of formal knowledge demonstration formalism which is utilized to symbolize the domain knowledge (Baader, Horrocks, & Sattler, 2008). The concepts of formal language along with the language characteristics of OWL are consequent from the description logics. A knowledge-based on description logic is constructed of two elements which are ABox and TBox (Baader et al., 2008). The knowledge base terminology in the TBox described and related to the concepts and also what they refer to. While the ABox comprises assertions around the individuals within the knowledge base. The description logic system architecture can be seen in 

·         Frames Paradigm: OWL also provides a surface syntax that is taken from the frames paradigm. Information regarding each class is grouped in Frames, so that ontologies become more simple to read and comprehend, especially for users that are unfamiliar with the  being used (Horrocks et al., 2010).

·         Resource Description Framework: To maintain maximum compatibility with existing web languages, the OWL ontology language extends the syntax and semantics of the Resource Description Framework adding numerous constructs and semantics intended for defining both the properties along with classes: interactions between classes, equality between others, cardinality, characteristics of properties, richer properties typing, as well as specified classes (Horridge, Knublauch, Rector, Stevens, & Wroe, 2004)

OWL sub-language of Knowledge Representation & Semantic Web Technologies

The OWL(Horridge et al., 2004) language is built up of the following three sub-languages, which offer dissimilar expressiveness levels along with the computation comprehensiveness:

·         OWL Lite: Defined as a simple sub-language used to support developers for constraints as well as classification hierarchy.

·         OWL DL: Description language to develop ontology. OWL DL can use automatic reasoner tools to check ontology’s consistency and build class hierarchy automatically. OWL DL is used in the state when maximum expressiveness is required.

·         OWL Full: recognized as the largely communicative sub-language, in case if OWL Lite and OWL DL are lacking in representing a complex ontology, OWL Full supports users where automatic reasoning is not needed.

OWL 2 sub-languages of Knowledge Representation & Semantic Web Technologies

OWL 2 (W3C, 2012) has received considerable attention leading to the development of tool support for the language. For example, Protégé has been extended to support the additional features provided by OWL 2, and also reasoners such as FaCT++and Pellet systems provide support for OWL 2 inference.

Computing the entire possible solutions of OWL 2 ontology might be challenging, and sometimes even turned out as an unsolved problem. To tackle this difficulty, OWL 2 has also come in the following sub-languages called profiles. These profiles are designed by applying syntactic boundaries on the full features of OWL 2 to simplify the reasoning process. The type option of profile to be used in practice depends on the reasoning task and the ontology structure.

·         OWL 2 EL: mainly suitable to be used for applications that need extremely great numbers of ontologies (properties and classes). OWL 2 EL provides expressive power for a large biomedical ontology.

·         OWL 2 QL: suitable for certain applications that have very huge data sample and need plenty of conjunctive queries using AND, OR, and NOT connectives. This profile is designed to store large data within a standard system of a relational database, and the database system rewrites an ontology query to an SQL query to get question answering.

·         OWL 2 RL:  suitable for applications that require scalable reasoning and expressive power at the same time. It is compatible with large instance data in the form of RDF triples, and a subset of OWL 2 might be applied by using . Reasoning with OWL 2 can be implemented by using rule-based reasoning engines. Reasoning and query answering issues could be conducted in a polynomial moment according to the size of the ontology.

·         OWL 2 RL/RDF:  different from the rest of the profiles. With OWL 2 RL/RDF we no longer talk about the syntactic restriction of OWL 2, but a semantic restriction of OWL 2 RDF-based Semantics. Users of OWL 2 can choose among two vaguely dissimilar semantics: TheDirect semantics and the RDF-based semantics. There is two way of assigning meaning to ontologies in OWL 2.

·         OWL 2 DL: under the direct semantics is decidable which makes designing a reasoner that answers all yes-no questions possible. It could also be expressed by using RDF-based semantics. In general, conclusions drawn using the direct semantics are still applicable under the RDF-based semantics while not all conclusions drawn by RDF-based semantics are valid under direct semantics as under the RDF-based semantics. There are some extra conclusions derived from viewing the ontologies as RDF graphs which can be regarded as a syntactically restricted version of OWL 2 Full which is designed to allow more RDF graphs or OWL 2full ontologies to be valid OWL 2 DL ontologies.

·         OWL 2 Full: another subset of OWL 2 which can only be interpreted under the semantics of RDF-based Semantics. Direct semantics allow OWL 2 ontologies to be expressed using Description Logic (DL), which is a section of first-order logic. RDF-based semantics defines as an RDFS semantics extension, grounded on observing OWL 2 ontologies as RDF graphs. The difference between direct semantic and RDF-based semantics is that Direct semantics does not apply to all RDF databases that utilize OWL characteristics, thus, it cannot be used with RDF graphs. OWL 2 Full under RDF-based semantics is unresolvable, therefore there are no such reasoners for OWL2Full under RDF-based semantics.

OWL2 Structure of Knowledge Representation & Semantic Web Technologies

Figure 3x OWL 2 language structure overview is designed with the intent to show its most important building blocks and how they connected one another. The top part of the diagram shows the various syntaxes that could be applied to switch ontologies.  Whilst the underneath part of the diagram shows the correspondence between the two semantic specifications (W3C, 2012).

The designed ontology model of this thesis will be using the OWL2 structure to coordinate education, health and social care interventions in the School Care Coordination system (SCCS).