A recommendation system using interviewers preferences for ranked candidate selection

Selection of an appropriate candidate among a group of applicants by multiple interviewers require decision making. The present work presents the design details of a recommendation system that generates a ranked list of selected candidates. The proposed recommendation system considers interviewers preferences and the sequence in which they wanted their preferred candidates to appear in the final selection list. However, finding an agreement (maximum satisfaction) among multiple preferences and variation in ordering for preferred candidates makes the problem more challenging. The proposed system models the input into score vector, interviewers' preference matrix and interviewers' satisfaction matrix to compute the final ranked candidate list. The Hungarian Aggregated Method (HAM) and the Greedy Aggregated Method (GRAM) are used for computing the final decision vector. The proposed system was evaluated using Fleiss's Kappa and prototype application. Results shows a better and confident recommendations by the proposed system.