A knowledge graph generally presents rather complex data connections for direct exploitation by a lambda user. The graphical interfaces developed over this type of graph must reduce the complexity of editing it and respond to certain criteria:
- - simplicity: the graphic interface must present a simple data display by grouping it according to various perspectives. This grouping allows the user to be guided during data exploration by presenting him, at each stage, with a partial view of the knowledge graph. Several grouping criteria can be imagined, the first of these being the exploitation of the data hierarchy that allows the display to be marked off at a single level for a given node on the graph. Groupings according to precise properties or classes can also be performed, in which case access to the data is granted following those criteria;
- - adaptability: the graphical interface must be automatically adapted to the ontology currently used to edit a knowledge graph and to all changes made on its classes and/or properties;
- - dynamism: the interface must be dynamically adapted to all changes occurring during the editing of a knowledge graph. This includes: changing the editing ontology and the selection of a new object to edit;
- - reusability: the interface must be reusable to edit different types of semantic graphs in the scope of any activity: media analysis, management of individuals, semantic search, management of corpuses, author publication.
Figure 4.8. Knowledge graph editing interfaces
The interfaces that we have developed to edit knowledge graphs respect these criteria and exploit all the completion functionalities presented in the previous parts of this work. They are each dedicated to a precise use, as presented in the table above (Figure 4.8).