Intelligent Analysis of Technical Support Requests in Service Desk Ticketing Systems

• Published in: 2020 International Conference on Engineering Management of Communication and Technology (EMCTECH) pp. 1 - 6
• Main Authors: Gorbunova, A. A., Trunov, A. S., Voronov, V. I.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 20-10-2020
• Subjects: actualizer system; API; application program interfaces; Atlassian Jira technical support system; classification; Convolutional codes; convolutional neural nets; convolutional neural network; convolutional neural networks; Dictionaries; input message text analysis; intelligent analysis; intelligent automatization; intelligent natural language text processing; Jira API; Kernel; message processing subsystem; natural language processing; network problem; request processing; Semantics; service desk ticket processing systems; spam filters; Task analysis; technical support requests; technical support services; text analysis; user requests; voice assistants; wireless network equipment; word embedding; word embedding approach; word2vec; work process
• DOI: 10.1109/EMCTECH49634.2020.9261539

Natural language processing methods are actively being used in systems that interact with people. These systems include voice assistants, spam filters, Service Desk ticket processing systems and many others. This article studies the problem of intelligent automatization of request processing. Requests are coming from users and describe a problem in wireless network equipment. The subsystem described in the article is integrated via specifically developed API to the Atlassian Jira technical support system to access user requests. Requests in technical support system contain information about a problem in the wireless network that the user encountered. The main goal of the subsystem is to analyze the input message, which is written in natural language. Methods of intelligent natural language text processing extract information about a network problem that can help locate a specific device and determine the cause of the problem, which can then be fixed by standard actions. A word embedding approach called word2vec is used for text analysis. The article describes the work process of convolutional neural networks and their integration with natural language processing methods for obtaining needed results, key modules that make up the subsystem, in particular, the message processing subsystem and the actualizer system and the relationship between them. Examples of input message text analysis and results obtained by the subsystem are shown.