6 Conclusions
Motivated by the need of efficiently managing large quantities of information in a human resources management system while still benefiting from novel non-standard reasoning services typical of knowledge representation and reasoning, we introduced a knowledge compilation approach in an originally designed relational schema, and devised solutions to execute inference services—both standard and non-standard ones—through standard-SQL queries only. We exploited such services referring to three relevant business processes typical of recruitment and human resources management, presenting them in the framework of the I.M.P.A.K.T. system. We reported an effective comparison with existing tools and research solutions and showed the effectiveness of our approach also from a computational point of view. Implementation of optimization techniques, such as table partitioning in our PostgreSQL database, are under development. As expected, first results show an improvement in the I.M.P.A.K.T. performance (e.g., for the skill matching execution over a dataset of 10000 profiles, we obtain a reduction of 30 percent on the retrieval time). Moreover, we are currently studying the peculiarities of the proposed design method for database modeling and management, with the aim of generalizing it to a framework fully independent from the underlying ontology.
Future work aims at testing further devised strategies for score calculation and at designing a service for CV translation from plain text according to our skill ontology. Moreover, in order to deal with specific business application requirements, e.g., the need to deploy I.M.P.A.K.T. in a more scalable environment, we are investigating the possibility of exploiting Big Data technologies for KB modeling and querying.
