Injecting domain knowledge in multi-objective optimization problems: A semantic approach

In the field of complex problem optimization with metaheuristics, semantics has been used for modeling different aspects, such as: characterization, parameters, decision-maker’s preferences, or algorithms. However, there is a lack approaches where ontologies are applied in direct way into process, aim enhancing it by allowing systematic incorporation additional domain knowledge. This due to high level abstraction ontologies, which makes them difficult be mapped code implementing problems and/or specific operators metaheuristics. this paper, we present strategy inject knowledge (by reusing existing creating new one) implementation that will optimized using metaheuristic. Thus, approach based on accepted enables building and exploiting computing systems problems. We describe methodology automatically induce user choices (taken from ontology) implementations provided jMetal framework. With illustrating our proposal, focus urban domain. Concretely, start defining an ontology representing city (e.g., building, bridges, point interest, routes, etc.) allows a-priori preferences decision maker standard, reusable, formal (logic-based) way. validate proposal several instances two use cases, consisting bi-objective formulations Traveling Salesman Problem (TSP) Radio Network Design (RND), both context scenario. The results experiments conducted show how semantic specification constraints effectively feasible solutions tackled TSP RND scenarios. aims at step forward towards automatic adaptation guided semantics, annotation human expert can now considered during process.