Research On Work Ticket Recommendation Methods For Intelligent Operation And Maintenance

Nowadays large scale IT service systems are becoming more and more complex,which brings high maintenance costs for handling failures.Thus,it is necessary to study intelligent maintenance methods,in order to provide high quality services.The problem report generated by operation and maintenance system is called ticket,which records problem details and solutions.Assigning tickets to the appropriate maintenance expert groups in time helps to improve the efficiency.The unstructured text and large number of domain words brings difficulties.Research on experts and routing recommendation is based on historical tickets,including the following aspects:1.In order to improve the accuracy for ticket recommendation,an algorithm based on convolutional neural network is proposed.The expert ability model is constructed based on the proficiency level and domain knowledge.Attention mechanism is introduced to study mutual influence between inputs.Then measures scores between tickets and expert models,and selects the top-k experts as the result.Experiments on real dataset show that,the accuracy is improved by about 6%,which verifies the feasibility of the proposed algorithm.The attention mechanism can effectively improve the feature extraction.2.Aiming at the rapid response requirement of ticket routing recommendation,a routing model based on expert collaboration network is proposed.The method first establishes an expert routing decision model by combining expert's professional knowledge and collaborative ability.A two-stage routing recommendation algorithm is proposed,including the initial expert recommendation and the selection strategy of the next candidate expert,and outputs routing sequence.The experimental results on real dataset show that,compared with historical routing decisions,the accuracy of the proposed algorithm reaches 96%,which verifies the effectiveness of the proposed algorithm.3.Based on the above research,a system of ticket management and analysis was developed.The system supports to import tickets from different sources.Including ticket management and statistics,data preprocessing to extract problem description information,ticket recommendation modules and a web-based visualization.It provides reliable management tools for administrators and helps to improve management efficiency.