Research And Application Of Automatic Classification Method For Work Tickets

With the increasingly standardized management of enterprises,most of its IT systems are constantly monitored and managed by a professional solution team.Any system problems or maintenance requests will be reported as a trouble ticket.Trouble tickets contain various detailed descriptions of the problems,understand these details,automatically classify trouble tickets and accurately predict the type of problem,that can help teams quickly find solutions,help service providers reduce system maintenance costs and improve service quality.Although trouble tickets have high application value,the details of problems are often hidden in complex information,difficult to be refined and contain rich semantic elements,which makes traditional trouble ticket classification methods difficult to accurately predict the type of problems.Because ontology fully considers the hierarchical relations and constraints between concepts,it can effectively solve the problem of mixed information and rich semantic elements in trouble tickets.Therefore,this paper studies the method of automatic classification of trouble tickets based on ontology to locate the details of the problem and improve the accuracy of trouble ticket classification as a goal to help enterprises reduce system maintenance costs.This paper studies the key technologies of trouble ticket classification,such as domain ontology,text representation model and classification algorithm.The main work is as follows:(1)We proposed a method based on LSA and clustering fusion of historical trouble ticket problem category anomaly detection method.This method uses the dimensionality reduction and denoising characteristics of LSA and the robustness of clustering fusion to detect the data of the problem types in the history trouble ticket.Experiments were conducted on trouble ticket data from real IT systems.The results show that the method can effectively improve the detection rate of abnormal rouble tickets in the problem category.(2)We constructed a trouble ticket domain ontology model.According to the characteristics of trouble tickets,combined with the general method of ontology construction,elaborated the construction process of the ontology model in the field of trouble tickets.Experiments on real data show that this method can effectively extract the knowledge in the trouble ticket field and efficiently construct the domain ontology model.(3)We proposed a method of automatic classification of trouble tickets based on domain ontology.Based on the ontology of trouble ticket,a reasonable trouble ticket reasoning model,concept vector space model and concept similarity measure algorithm are designed.The K-nearest neighbor classifier is further used to classify the trouble ticket automatically with the trouble ticket domain ontology model.Experiments on real data show that this method can effectively improve the accuracy of trouble ticket classification.