**7. Knowledge graphs enrichment: semantic annotation**

The database world is a place that is controlled by computers. Supercomputers have amazing computing capabilities, but they can be a struggle when it comes to acquiring new knowledge and experience or putting knowledge into practice. While it is easy for a human to decide whether two or more things are related based on cognitive associations, a computer often fails to do it. Unlike traditional lexical search where search engines look for literal matches of the query words and their variants, semantic annotation tries to interpret natural language close to how people do it. During semantic annotation, all references to cases related to entities in the ontology are recognized. Semantic annotation is the glue that ties ontologies into document spaces, via metadata.

The working panel for implementing the semantic annotation process is shown in **Figure 9** below. At the top of this panel is a workspace for entering and editing network resource addresses (URLs) to be annotated. The data in this workspace can be entered from any source, including manually. However, a more technologically advanced approach is to first find on the WWW those network resources that are most adequate to a given domain using the "Context–sensitive search" software agent. The found adequate content can then be easily loaded using the "Download resources" button and included in the list for annotation with a single mouse click.

The settings panel for the semantic annotation process is shown in **Figure 10** below. For annotation, you can select any of the knowledge graphs that are presented in the semantic repository, as well as any combination of them. To calculate measures of similarity between the annotated resource and entities from knowledge graphs, both text analysis methods and neural networks that are trained on existing knowledge graphs can be used.

#### **Figure 10.**

*Setting the options for the semantic annotation process: 1 – selecting and visualizing the knowledge graphs used; 2 – selecting of the technology and setting semantic annotation parameters.*

Let us define the function *<sup>C</sup>D*ð Þ <sup>Ω</sup> for selecting the best snippets as follows:

Here, the value *D*<sup>Ω</sup> ¼ min *<sup>x</sup>*<sup>∈</sup> <sup>Ω</sup> *D*Ωð Þ *x* is called the index of dominance of the whole Ω set. Snippets with a minimum value of the dominance index form the Pareto set. The Pareto set includes snippets that are the best with respect to all the

In the project [4], an intuitively more acceptable value is used as the index of dominance, equal to the difference between the number of aspects taken into account and the dominance index determined by the formula (Eq. (4)). Groups of snippets with the same value of the dominance index form clusters, which in the final output of the "Context–sensitive search" software agent are arranged in

As an illustration of the previous computations in the next section **Figure 9** shows a variant of sorting snippets by dominance index. Snippets are sorted in descending order of the dominance index value when six metrics are taken into account, including snippets relevance and pertinence. When snippets are ordered by the value of the dominance index, within groups of elements with the same value of the dominance index (that is, within a cluster), the snippets are ordered by each of the metrics taken into account in the calculations. Other ways to organize and systematize the content found are available for any combination of metrics that

*Selecting network resources for semantic annotation: 1 – workspace for entering and editing network addresses (URLs) to be annotated; 2 – setting options and loading results of the context-sensitive search; 3 – the most*

*z* ∈ Ω

*D*Ωð Þ*z*

(5)

*<sup>C</sup>D*ð Þ¼ <sup>Ω</sup> *<sup>x</sup>*<sup>∈</sup> <sup>Ω</sup> : *<sup>D</sup>*Ωð Þ¼ *<sup>x</sup>* min

considered aspects, including relevance and pertinence.

*Cloud Computing Security - Concepts and Practice*

descending order of this index.

**Figure 9.**

**110**

*relevant results of the context-sensitive search.*

characterize the adequacy of the snippets.

theory development and technology's implementation for semantic web, description logics and incarnations of the ontologies description language OWL. A recent qualified review [11] gives a fairly complete picture of the progress made in this

*Semantic Web and Interactive Knowledge Graphs as an Educational Technology*

the statistic computations, using aggregations over the hierarchy levels.

In contrast to the above solutions, the project [4] is mainly focused on the implementation in educational activities of universities and is not limited to visualization of knowledge graphs and interactive navigation, but is aimed at the introduction of the latest semantic web technologies to the training process, taking into account the achievements in the field of uncertain reasoning. The results obtained and the software created are used in the real educational process at National Research Nuclear University MEPhI, and the project, as a whole, is focused on the practical mastering of semantic web technologies by students and professors.

The reported study was funded by the Russian Foundation for Basic Research and Government of the Kaluga Region according to the research projects 19-47- 400002 and was funded by the Vladimir Potanin Foundation according to the

Special mention should be made on the project [12], where for the first time an attempt was made to put into practice the methods of inductive reasoning for the purpose of semantic annotation of content from the WWW. As for the issues of visualization linked data [13], here, one of the first successful projects was Lodlive [14], which provided a tool for easier surfing through the DBpedia knowledge database. It is important to continue to develop and improve tools for intuitive perception of linked data for non-professionals. VOWL [15] is one of the modern project for the user-oriented representation of ontologies; it proposes the visual language, which is based on a set of graphical primitives and an abstract color scheme. As noted in [3], LinkDaViz [16] proposes a web-based implementation of workflow that guides users through the process of creating visualizations by automatically categorizing and binding data to visualization parameters. The approach is based on a heuristic analysis of the structure of the input data and a visualization model facilitating the binding between data and visualization options. SynopsViz [17] is a tool for scalable multilevel charting and visual exploration of very large RDF & Linked Data datasets. The adopted hierarchical model provides effective information abstraction and summarization. Also, it allows to efficiently perform

area and the directions for further research.

*DOI: http://dx.doi.org/10.5772/intechopen.92433*

**Acknowledgements**

project GC190001383.

**113**

#### **Figure 11.**

*Displaying semantic annotation results: 1 – addresses of the annotated network resources (URLs); 2 – setting options and starting the semantic annotation process; 3 – network resource for which semantic annotation is performed; 4 – knowledge graphs and entities corresponding to the annotated network resource.*

It is possible to annotate network resources using classes (concepts) of the corresponding ontology (TBox – terminological components), using objects (individuals) of knowledge graphs (ABox – assertion components), or using both.

The depth of the carried out semantic analysis can be limited by considering only textual metadata inherent in network resources and entities in knowledge graphs. Full-text semantic analysis can be very expensive and, in many ways, redundant. Improving the accuracy of annotation in full-text analysis often does not justify the increased consumption of computing resources and time.

The number of displaying entities from knowledge graphs can be limited by the user. At the top of the output of the "Semantic annotation" software agent, the entities that are most adequate to the annotated resource appear. All the results of the work can be saved in files on the user's computer for later study.

As an example of using the "Semantic annotation" software agent, **Figure 11** below shows the results of the semantic annotation of one network resource. It can be seen that semantic annotations from five different knowledge graphs were discovered. With one click, the user can open the RDF browser and visualize the found annotations in any of the knowledge graphs, as well as anyone can see the surroundings of the entities found, for example, their classes and neighboring objects. This information is essential for a knowledge engineer who is engaged in knowledge graph refinement and enrichment.

### **8. Related work and conclusion**

Groups of scientists from the University of Manchester, Stanford University, University of Bari and a number of other universities are focused on the issues of

### *Semantic Web and Interactive Knowledge Graphs as an Educational Technology DOI: http://dx.doi.org/10.5772/intechopen.92433*

theory development and technology's implementation for semantic web, description logics and incarnations of the ontologies description language OWL. A recent qualified review [11] gives a fairly complete picture of the progress made in this area and the directions for further research.

Special mention should be made on the project [12], where for the first time an attempt was made to put into practice the methods of inductive reasoning for the purpose of semantic annotation of content from the WWW. As for the issues of visualization linked data [13], here, one of the first successful projects was Lodlive [14], which provided a tool for easier surfing through the DBpedia knowledge database. It is important to continue to develop and improve tools for intuitive perception of linked data for non-professionals. VOWL [15] is one of the modern project for the user-oriented representation of ontologies; it proposes the visual language, which is based on a set of graphical primitives and an abstract color scheme. As noted in [3], LinkDaViz [16] proposes a web-based implementation of workflow that guides users through the process of creating visualizations by automatically categorizing and binding data to visualization parameters. The approach is based on a heuristic analysis of the structure of the input data and a visualization model facilitating the binding between data and visualization options. SynopsViz [17] is a tool for scalable multilevel charting and visual exploration of very large RDF & Linked Data datasets. The adopted hierarchical model provides effective information abstraction and summarization. Also, it allows to efficiently perform the statistic computations, using aggregations over the hierarchy levels.

In contrast to the above solutions, the project [4] is mainly focused on the implementation in educational activities of universities and is not limited to visualization of knowledge graphs and interactive navigation, but is aimed at the introduction of the latest semantic web technologies to the training process, taking into account the achievements in the field of uncertain reasoning. The results obtained and the software created are used in the real educational process at National Research Nuclear University MEPhI, and the project, as a whole, is focused on the practical mastering of semantic web technologies by students and professors.
