last paragraph of future work, users in the center of the process

8_conclusions/conclusions.tex

@@ -24,5 +24,7 @@ \section{Future Work}
 
 One of the main future lines we have identify during this thesis, extending the contributions on the enhancement of KGC systems, is to define methodologies and techniques for an optimal physical design of knowledge graphs. The main idea is to be able to decide which parts of a KG have to be materialized or virtualized analyzing the features of the typical inputs of a KGC process (data, constraints, mapping rules, ontology). We believe that these methodologies will help to start to see the web as an integrated database that can be queried using Semantic Web technologies. Together with the application of the optimizations techniques proposed in this thesis over distributed environments, such as the ones proposed in \citep{endris2019ontario,mami2019squerall}, leverage the use of declarative KGC techniques to its next steps providing the basis for developing real-world knowledge graph applications.
 
-For the evaluation systems, we need to extend the current proposals in order to coverage other be more flexibly to evaluate a KGC workflow, taking into account all the parameters that can have an impact in their behavior. Some examples of these possible future lines are: the inclusion of mapping rules with transformation functions, the adaption of mapping rules construction in a data integration system to isolated parameters from this input in the evaluation, or the improvement in creation of datasets at scale, exploiting the information from mapping rules or graph constraints (SHACL). Finally, it is important to create systems that include a ground truth in order to test not only the performance and scalabitily of the engines but also other important features such as correctness and completeness.
+For the evaluation systems, we need to extend the current proposals in order to coverage other be more flexibly to evaluate a KGC workflow, taking into account all the parameters that can have an impact in their behavior. Some examples of these possible future lines are: the inclusion of mapping rules with transformation functions, the adaption of mapping rules construction in a data integration system to isolated parameters from this input in the evaluation, or the improvement in creation of datasets at scale, exploiting the information from mapping rules or graph constraints (e.g, SHACL shapes). Finally, it is important to create evaluation systems that include a ground truth in order to test not only the performance and scalabitily of the engines but also other important features such as correctness and completeness.
+
+Finally, the use of declarative and standard mapping rules and metadata descriptions makes possible the generalization of KGC engines and optimizations, avoiding ad-hoc and manual steps. It also incorporates a set of important benefits for these processes such as the improvement of its maintainability, readability, and understandability. We believe that this kind of solutions should be promoted in academic, industry and public organizations as good practices for data management on the web to for example, avoid to have data cemeteries such as the current open data portals. Our vision is that, analyzing the role of the users in complex data integration environments on the web, will help to understand how to promote and develop robust and useful semantic web solutions for constructing knowledge graphs at scale in distributing scenarios.