**3.1 Architecture of the SBLODF framework**

*Linked Open Data - Applications, Trends and Future Developments*

facts, *A*, [11]. Thus;

respectively.

domains in question [43].

following section.

follows:

or schema, it means that the processes can be represented in the form of classes, relations, individuals, and axioms (C,R,I,A). Thus, we note that the structural layer of ontologies can be defined as a *quadruple* which are construed on connected sets of taxonomies (RDF + Axioms) or yet formal structure (Triple + Facts). Whereby the *subjects* include the represented class(es), *C*, the *objects* include the individual process elements or instances, *I*, the *predicates* are used to express the relationships, *R*, that exist amongst the subjects and objects, and then sets of axioms that state

Following the aforementioned definition of the ontological concept or schema, this work note that ontologies serve and are built to perform the main functional mechanisms for the integration of data models for the various systems (e.g. LOD) as

• *Conceptualisation:* method used to represent abstract models of a phenomenon in real-world settings. This is done by identifying suitable domain (semantic) relationships that exist amidst the process elements (concepts) through formal definitions in what can be called declarative axioms that allow for the resultant

• *Explicitness:* procedures that allow or support the different types of concepts and restrictions on their use (properties assertions) to be defined explicitly.

• *Formality:* expressions which are defined to prevent unexpected interpretation of the C,R,I,A as quadruple (e.g. concepts and notations, relationships, properties restrictions, etc.). Thus, it enables the resultant systems or models to be machine-readable and machine-understandable,

The representation (modeling) of knowledge using ontologies (e.g. taxonomies) helps in organizing *metadata* for complex information or data structures. According to Sheth et al. [41], description of real-time processes through metadata creation provides a syntactic as well as semantic way of representing information about the resources that are encoded as instances (entities) in ontological form. Besides, the formal representation of ontologies and the underlying metadata created as a result of the representations allows for automatic reasoning of the processes by making references (inference) to the defined concepts [42]. Indeed, with such reasoning aptitude, the process analysts or owners are able to ensure specification of the process domains (knowledge) in view in an ontological form that can logically be interpreted in an apt way. Consequently, this permits for automatic reasoning of the different concepts to derive an explicit/implicit knowledge about the process

Therefore, the main benefits of ontologies for formal integration of datasets and models in any shape or platform can be summarized in two forms: (i) encoding knowledge about the specific process domains, and (ii) conceptual analysis and reasoning of the processes at more abstraction levels as described in detail in the

models to be represented (conceptualisation) declaratively.

**3. Proposed semantic-based LOD framework (SBLODF)**

Ont = (C,R,I,A) (1)

**44**

Information retrieval and structuring of the different sets of data that are stored in several databases or knowledge-base are usually performed in alignment with the users' query [38]. As gathered in **Figure 1**, the supported formats may be a list of document files or keywords issued to the system through the query module (functional operators). In turn, the retrieval module references the properties descriptions (conceptual assertions) that underlie the (semantic) models to produce information that is relevant to the users' query. For example, using the superClass-subClass hierarchies that are usually defined in a taxonomical form in ontologies. This is done through the classification process (e.g. classifying by using a reasoner) to compute the relevant information (e.g. individual entities or process instances) that fulfills the properties restriction by definition [44]. Technically, the most fitting (related) concepts are then presented to the user in a formal way, e.g. explicitly and implicitly.

Furthermore, we note that information retrieval and extraction systems such as the SBLODF framework (**Figure 1**) typically do not only support unstructured data or documents (e.g. textual data), but it also deals with semi-structured and structured data. This is where the semantic technologies and such type of systems (which combines the information retrieval (IR) with information extraction (IE) features) [38] becomes greatly beneficial. Functionally, the resulting system allows for merging and manipulation of structured, semi-structured, and unstructured data through the search (query) modules by enabling a conceptual intersection or reasoning between the different elements as contained in the system. Thus, the SBLODF is referred to as a conceptualization method or information processing system that combines the features of the machine-readable and machine-understandable systems or mechanisms.

For example, enterprise vendors such as FAST (a Microsoft subsidiary) incorporated analytical search functions to support data visualization and reporting into

#### **Figure 1.** *Semantic-based linked open data framework (SBLODF).*

their products [38, 45]. Moreover, Ingvaldsen [38] notes that the business process intelligence (BPI) solutions and offerings can also benefit from such a combination (IR and IE supported systems) by giving the users a search facilitated (data analysis) environment to harvest/harness data from both structured and unstructured data sources. Thus far, giving the users a more flexible environment for accessing relevant data items.

Interestingly, semantic-based information retrieval and extraction systems as illustrated in **Figure 1**, represents to be a step further in supporting the BPI's by providing additional modules or components that allows for integrating metadata description (e.g. ontologies) to the system design or framework. The semantic-based components (see: **Figure 1**) aims to add a machine tractable and/or re-purposeable layer of annotations that are relative to ontologies in order to complement the existing web of information and data analysis procedures, or yet, the omnipresence of natural language hypertext [4, 46, 47]. Perhaps, this is fundamentally done through the creation of semantic annotations [11, 23] and linking of the different concepts or modules to ontologies. In turn, the semantically motivated process or models turns out to become automatic or semi-automatic in nature and allows for ample integration of the LOD frameworks due to creation, interrelation, or application of the ontologies (semantic schema). Besides, this has led to the advancement of hybrid intelligent systems such as the ontology-based information extraction systems (OBIE) [9, 13, 15]. Explaining why IE and semantic technologies can be used to bring together a common language or syntax upon which the LOD systems or web search are built specially given the ever-needed formal knowledge or tools for information (data) access and utilization.

Some examples of state-of-the-art tools or systems that trails to support the semantic-based LOD framework or search include; KIM (knowledge and information management system) [31, 48] an extendable platform for information management that seemingly offers IE-based functions for metadata creation and search. Technically, KIM consists of a set of front-end (user-interface) for online information search by offering semantically-enhanced browsing features.

Another tool that tends to support the semantic-based LOD, such as the SBLODF framework described in this chapter, is Magpie [49]. Magpie is developed and implemented as an add-on to web browsers by using IE mechanisms to support collaborative information interpretation and modeling of the extracted knowledge from the web. As illustrated in **Figure 1**, it annotates the different web pages with metadata descriptions in an automated manner by automatically populating ontologies from the relevant (web) sources. Thus, the application (Magpie) is interoperable with ontologies or semantic schema. Moreover, it is important to mention that one of the fundamental elements of the tool (Magpie) that is pertinent to this work is the fact that it makes use of ontologies to provide specific (tailored content) information to the users.

There are several other platforms that can be referred to also support the SBLODF framework. This includes the SemTag [50] which utilizes IE facilities or function to support large scale semantic annotations and process descriptions using TAP ontology. As described in **Figure 1**, SemTag functions by performing annotation of all defined mentions (references) of any given process instance or entity in the ontology (TAP) through a lookup phase. This lookup process is then followed by the disambiguation phase during which it assigns the right classes (or establishes instances that do not correspond to a class in the TAP) using a vector-space model [50].

**47**

**Figure 2.**

*procedure [11].*

*Linked Open Data: State-of-the-Art Mechanisms and Conceptual Framework*

**4. Implementation components of the semantic-based linked open data** 

The work describes in this section (**Figure 2**) how the semantic schema is used to support the development of the LOD framework. Ontology-based information retrieval and extraction systems such as the SBLODF (**Figure 1**) are construed on

• *Named Entity recognition* (NE) which trails to find and classifies the different

• *Co-reference resolution* (CO) which identifies the relations or association that

• *Template Element construction* (TE) that adds descriptive information (meta-

• *Template Relation construction* (TR) that locates the links or references between

• *Scenario Template production* (ST) that matches (fits) the TE and TR compo-

*Implementing the semantics components in SBLODF using create-link-map-check-use (CLMCU)* 

concepts that can be found within the model or knowledge-base.

*DOI: http://dx.doi.org/10.5772/intechopen.94504*

co-exist amongst the concepts or entities.

data) to the classified NE through the CO component.

nents into a specified scenario or process instance.

**framework (SBLODF)**

the main building blocks [31]:

the TE (entities), and
