SNOOKER: a dataset generator for helpdesk services

Abstract

The availability of suitable datasets and data generators is crucial for developing intelligent systems, especially in helpdesk services. However, the lack of publicly accessible data generators focused on helpdesk operations, where incidents are often closed without detailing the treatment procedures, poses challenges to implementing intelligent systems such as recommender systems. To address this issue, a dataset generator can be employed to simulate helpdesk incidents. This paper introduces SNOOKER (dataSet geNeratOr fOr helpdesK sERvices), a customizable dataset generator designed to create and treat helpdesk tickets, including domain-specific incidents (e.g., cybersecurity) by orchestrating simulated actions and multiple IT teams. SNOOKER’s output is compared against a real anonymized dataset from S21Sec Cyber Solutions by Thales. The datasets are evaluated using Kolmogorov-Smirnov, Kullback-Leibler Divergence, and Hellinger distance tests, with results indicating similar distributions. For example, the first metric returned a low K-S value and a p-value exceeding 5%, while the second and third measures presented 0.003 and 0.03, respectively. Furthermore, experiments with different team configurations revealed that ticket scheduling highly depends on each team’s operators’ numbers and work shifts, increasing with unbalanced shifts and fewer operators.