A Text Mining-Based Decision Support System for Candidate Selection in Government Recruitment Processes

Abstract

Long processing times, manual labor, and human biases are just a few of the major inefficiencies that frequently plague the public sector hiring process and cause delays in the timely acquisition of qualified talent. In order to improve and automate the hiring process for government agencies, this paper suggests a brand-new text mining-based decision support system. The suggested system parses, examines, and rates resumes in accordance with particular job descriptions using machine learning and natural language processing (NLP) techniques like XGBoost Classifier, TF-IDF, and Sentence-BERT Embeddings. The results demonstrate that the candidate selection decision support system successfully integrates structured features and text-based semantic analysis, improving candidate ranking and reducing false positives and negatives while encouraging ethical recruitment practices. Its accuracy, precision, and recall for identifying relevant candidates within the top 20% of the applicant pool are 82%, 92%, and 85%, respectively.