Business process retrieval based on behavioral semantics

This paper develops a framework for retrieving business processes considering search requirements based on behavioral semantics properties; it presents a framework called “BeMantics” for retrieving business processes based on structural, linguistics, and behavioral semantics properties. The relevance of the framework is evaluated retrieving business processes from a repository, and collecting a set of relevant business processes manually issued by human judges. The “BeMantics” framework scored high precision values (0.717) but low recall values (0.558), which implies that even when the framework avoided false negatives, it prone to false positives. The highest precision value was scored in the linguistic criterion showing that using semantic inference in the tasks comparison allowed to reduce around 23.6 % the number of false positives. Using semantic inference to compare tasks of business processes can improve the precision; but if the ontologies are from narrow and specific domains, they limit the semantic expressiveness obtained with ontologies from more general domains. Regarding the performance, it can be improved by using a filter phase which indexes business processes taking into account behavioral semantics properties.