Manual categorization of service desk tickets may lead to dispatching of problem tickets to aninappropriateexpertgroup,delays the response time and interrupts the normal functioning of the business.Traditional machine learningapproachescan be applied to train an automated service desk ticket classifier by mining the historical tickets.Sparsity, non-linearity, overfitting and handcrafting of features are some of the issues concerning the traditional ticket classifiers.This research work proposes a deep neural network model based on Convolution Neural Network (CNN) for the automated classification of service desk tickets.A real-world service desk ticket data is used to corroborate the efficiency of the proposed ticket classifier model and compared the results with the traditionalclassifiers.The proposed CNN model outperforms the chosen alternatives in terms of overall model performance.Benefits of the proposed model includes assignment of tickets to the correct domain groups,speedy resolution, improved productivity, uninterrupted business and customer satisfaction.

Service desk systems, Machine learning, Deep learning, Convolution Neural Network, Word Embeddings, Text classification, Feature maps.