[4Yin2] Interactive session 2

[4Yin2-43] Development of similar human resources search function for human resource management using UMAP

〇Takehiro Ohashi¹, Tsutomu Ishida¹ (1.Fujitsu Ltd.)

Keywords:Human Resource Management, Similarity Search

In the domestic IT industry, there is a problem of organizational silos with subdivided and multi-stage group company structures and multi-layer subcontracting structures, and it is not possible to flexibly allocate human resources due to the isolation of functions within the organization, which impedes sharing of knowledge and skills. As a solution to the problem, simple matching between a project and human resources was performed with a resource management system. However, when matching the skill set required for a project with the skill set of personnel, requests are concentrated on specific personnel that leads to the consequence of finding a substitute. In that case, finding personnel with similar skill sets leads to resource optimization. In this study, we used Uniform Manifold Approximation and Projection (UMAP), a nonlinear dimensionality reduction algorithm, to develop a similar human resource searching functionality for a wide variety of human resources data. The embedding of similar human resources in low-dimensional space was obtained by UMAP with cosine distance as the metric, and k nearest neighbor (kNN) search is used to obtain similar human resources and similarity scores from applicable personnel. This function made it possible to find a substitute in a short time and contributed to the pinpoint placement of appropriate human resources.