|کد مقاله||کد نشریه||سال انتشار||مقاله انگلیسی||ترجمه فارسی||نسخه تمام متن|
|4950259||1364283||2018||18 صفحه PDF||ندارد||دانلود کنید|
â¢A trouble ticket classification framework using ticket partition and signature construction is presented.â¢The ticket partition and signature construction is regarded as an optimization problem and is solved by a local search strategy.â¢A signature based ticket classification algorithm is proposed to identify the problem type of an incoming ticket.â¢An empirical study on real world ticket data from a large enterprise IT infrastructure is conducted.
When a critical system exhibits an incident during its operation, a ticket is usually generated by the monitoring systems or users to describe its issue and should be fixed by system maintenance teams in an acceptable short period of time to avoid serious economic or reputation losses. Although there are a few works about ticket classification, they suffer from poor performance because of the obvious characteristics of unstructured, short free-text with large vocabulary size, large volume, and so on.To address this performance issue, this paper proposes a trouble ticket classification framework that automatically and accurately identifies the problem type of an incoming ticket. First, a ticket partition and signature construction algorithm is developed, which integrates domain knowledge to improve the quality of data preparation and applies a local search strategy to simultaneously construct ticket groups and their signatures. And then, a signature based ticket classification algorithm is proposed to identify the problem type of an incoming ticket by finding a group signature with the most similarity satisfying the similarity threshold. To demonstrate the effectiveness of the proposed solution, we empirically validate it on real world ticket data from a large enterprise IT infrastructure. Experiments show that our solution outperforms other alternatives in terms of the overall performance.
Journal: Future Generation Computer Systems - Volume 78, Part 1, January 2018, Pages 41-58