Building Relevant Electronic Profiling for Automated Career Recommendations

Abstract:

Recommender systems are information filtering systems that offer suggestions from a vast variety and quantity of objects which are of user’s interest. As a result, the so-called user is guided in a unique way to useful or interesting objects in a large space of possible options. Without any doubt, the structure of the user model in such systems has a major influence on the success of recommendations. Widely used in e-commerce, recommender systems are more and more present in other domains as well. The current paper proposes a methodology to build relevant e-profiles in smart career path recommender systems, based on a survey filled in by high school and university students related to their perceived usefulness of various data sources necessary for the automatic implementation of career guidance process. Our methodology provides, besides steps to build the connectors to data sources, the process of choosing the relevant features used to develop both static and dynamic profiling, the possible recommendation techniques for those features and the challenges in applying the methodology. When discussing the issue of profiling in career recommender systems, we could take into consideration the user profile and the job profile, the recommendation mechanism trying to find the optimal match between those two. Our methodology is applicable for building both profiles, among the techniques we used to gather data are crowdsourcing, social network interrogation, CV parsing.