IP Library Granted Patent US 12,657,532
Granted Patent B2
US 12,657,532 · App. 17/548,768 · Granted Jun 16, 2026

Ticket embedding based on multi-dimensional IT data

Inventors: Zhi Wang (ShangHai, CN); Zhao Qi Wu (Shanghai, CN); Li Na Yuan (Beijing, CN); Qian Ke Fang (Ningbo, CN); Li Long Chen (Beijing, CN)
Assignee: International Business Machines Corporation
G06Q10/06311G06N3/04G06N3/08
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,657,532
App. No.
17/548,768
Granted
Jun 16, 2026
Kind
B2
Abstract

An approach is provided that trains an artificial intelligence (AI) system, such as a neural network, to process IT ticket data. The approach receives IT tickets from various ticket sources. Ticket vectors corresponding to each of the IT tickets are computed. An analysis is performed using the ticket vectors and a node vector that corresponds to a network topology. The analysis is performed using a corpus of IT ticket data. An IT ticket model used by the AI system is trained based on the analysis. Responses are provided to requestors of the AI system using the trained IT ticket model.

Claims (64)

1 . A computer-implemented method, implemented by an information handling system that includes a processor and a memory, the method comprising:

receiving a plurality of information technology (IT) tickets from one or more ticket sources;

calculating a topology graph comprising an edge weight of nodes, wherein the calculating comprises randomly discovering a relationship among connected nodes, and generating sampled series data describing a topology structure of a network, and wherein the edge weight is a sum of frequency of a ticket appearing in two adjacent nodes, based on one or more input parameters;

pre-processing the plurality of IT tickets using Doc2Vec to embed descriptive information of the IT tickets to generate a word-level vector;

using self-attention learning, merging the word-level vectors, outputting a weighted vector as a hidden input layer to train a neural network, until a loss gap between a softmax function and the hidden input layer is below a defined threshold; and

based on the loss gap between the softmax function and the hidden input layer being below the defined threshold, providing by a trained model, responses to one or more requestors of an artificial intelligence (AI) system using the trained IT ticket model.

2 . The method of claim 1 , wherein each of the ticket vectors includes the ticket vector and a side information vector, the method further comprising: gathering a set of side information corresponding to each of the IT tickets; and processing the gathered side information into the side information vector.

3 . The method of claim 2 wherein the side information includes at least one data item selected from the group consisting of a ticket comment, a ticket work note, a server location, a geographic location, and a customer identifier.

4 . The method of claim 2 further comprising:

selecting the ticket vector, the node vector, and the side information vector corresponding to a selected IT ticket; and

merging the selected ticket vector, the selected node vector, and the selected side information vector, the merging resulting in a set of hidden layer data that is utilized to train a neural network in the AI system.

5 . The method of claim 4 further comprising:

performing a self-attention process on the selected ticket vector, the selected node vector, and the selected side information vector, the self-attention process resulting in a weighted vector utilized to train the neural network.

6 . The method of claim 5 further comprising:

calculating a difference between a result of a softmax function performed on the weighted vector and the set of hidden layer data; and

repeatedly performing the self-attention process and the calculating until the difference is below a given threshold.

7 . The method of claim 1 further comprising:

generating the node vector by randomly traversing through a selected plurality of nodes from a set of sampled series data that describes a topology structure of a network;

calculating a weight between the selected plurality of nodes; and

utilizing a sampling strategy to describe a frequency of transitions between the selected plurality of nodes.

8 . An information handling system comprising:

one or more processors;

a memory coupled to at least one of the processors;

a set of computer program instructions stored in the memory and executed by at least one of the processors in order to perform actions comprising:

receiving a plurality of information technology (IT) tickets from one or more ticket sources;

calculating a topology graph comprising an edge weight of nodes, wherein the calculating comprises randomly discovering a relationship among connected nodes, and generating sampled series data describing a topology structure of a network, and wherein the edge weight is a sum of frequency of a ticket appearing in two adjacent nodes, based on one or more input parameters;

pre-processing the plurality of IT tickets using Doc2Vec to embed descriptive information of the IT tickets to generate a word-level vector;

using self-attention learning, merging the word-level vectors, outputting a weighted vector as a hidden input layer to train a neural network, until a loss gap between a softmax function and the hidden input layer is below a defined threshold; and

based on the loss gap between the softmax function and the hidden input layer being below the defined threshold, providing by a trained model, responses to one or more requestors of an artificial intelligence (AI) system using the trained IT ticket model.

9 . The information handling system of claim 8 , wherein each of the ticket vectors includes the ticket vector and a side information vector, wherein the actions further comprise:

gathering a set of side information corresponding to each of the IT tickets; and

processing the gathered side information into the side information vector.

10 . The information handling system of claim 9 wherein the side information includes at least one data item selected from the group consisting of a ticket comment, a ticket work note, a server location, a geographic location, and a customer identifier.

11 . The information handling system of claim 9 wherein the actions further comprise:

selecting the ticket vector, the node vector, and the side information vector corresponding to a selected IT ticket; and

merging the selected ticket vector, the selected node vector, and the selected side information vector, the merging resulting in a set of hidden layer data that is utilized to train a neural network in the AI system.

12 . The information handling system of claim 11 wherein the actions further comprise:

performing a self-attention process on the selected ticket vector, the selected node vector, and the selected side information vector, the self-attention process resulting in a weighted vector utilized to train the neural network.

13 . The information handling system of claim 12 wherein the actions further comprise:

calculating a difference between a result of a softmax function performed on the weighted vector and the set of hidden layer data; and

repeatedly performing the self-attention process and the calculating until the difference is below a given threshold.

14 . The information handling system of claim 8 wherein the actions further comprise:

generating the node vector by randomly traversing through a selected plurality of nodes from a set of sampled series data that describes a topology structure of a network;

calculating a weight between the selected plurality of nodes; and

utilizing a sampling strategy to describe a frequency of transitions between the selected plurality of nodes.

15 . A computer program product stored in a computer readable storage medium, comprising computer program code that, when executed by an information handling system, performs actions comprising:

receiving a plurality of information technology (IT) tickets from one or more ticket sources;

calculating a topology graph comprising an edge weight of nodes, wherein the calculating comprises randomly discovering a relationship among connected nodes, and generating sampled series data describing a topology structure of a network, and wherein the edge weight is a sum of frequency of a ticket appearing in two adjacent nodes, based on one or more input parameters;

pre-processing the plurality of IT tickets using Doc2Vec to embed descriptive information of the IT tickets to generate a word-level vector;

using self-attention learning, merging the word-level vectors, outputting a weighted vector as a hidden input layer to train a neural network, until a loss gap between a softmax function and the hidden input layer is below a defined threshold; and

based on the loss gap between the softmax function and the hidden input layer being below the defined threshold, providing by a trained model, responses to one or more requestors of an artificial intelligence (AI) system using the trained IT ticket model.

16 . The computer program product of claim 15 , wherein each of the ticket vectors includes the ticket vector and a side information vector, wherein the actions further comprise:

gathering a set of side information corresponding to each of the IT tickets; and

processing the gathered side information into the side information vector.

17 . The computer program product of claim 16 wherein the side information includes at least one data item selected from the group consisting of a ticket comment, a ticket work note, a server location, a geographic location, and a customer identifier.

18 . The computer program product of claim 16 wherein the actions further comprise:

selecting the ticket vector, the node vector, and the side information vector corresponding to a selected IT ticket; and

merging the selected ticket vector, the selected node vector, and the selected side information vector, the merging resulting in a set of hidden layer data that is utilized to train a neural network in the AI system.

19 . The computer program product of claim 18 wherein the actions further comprise:

performing a self-attention process on the selected ticket vector, the selected node vector, and the selected side information vector, the self-attention process resulting in a weighted vector utilized to train the neural network.

20 . The computer program product of claim 15 wherein the actions further comprise:

generating the node vector by randomly traversing through a selected plurality of nodes from a set of sampled series data that describes a topology structure of a network;

calculating a weight between the selected plurality of nodes; and

utilizing a sampling strategy to describe a frequency of transitions between the selected plurality of nodes.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Dec 13, 2021
From: WANG, ZHI; WU, ZHAO QI; YUAN, LI NA; FANG, QIAN KE; CHEN, LI LONG
To: INTERNATIONAL BUSINESS MACHINES CORPORATION
Reel/Frame 058369/0270 →
Continuity (1)
Related Publication 20230186190A1 · Jun 15, 2023
References Cited (30)
US 9171102B1 · Zlatnik · 2015 [cited by examiner]
US 10740692B2 · Mann · 2020 [cited by examiner]
US 11803402B1 · Gupta · 2023 [cited by examiner]
US 11954609B2 · Nagarajan · 2024 [cited by examiner]
US 20070058564A1 · Agrawala · 2007 [cited by examiner]
US 20070076601A1 · Wang · 2007 [cited by examiner]
US 20140207385A1 · Martin · 2014 [cited by examiner]
US 20190318204A1 · Mishra · 2019 [cited by examiner]
US 20190347282A1 · Cai · 2019 [cited by applicant]
US 20200104752A1 · Meng · 2020 [cited by examiner]
US 20200293946A1 · Sachan · 2020 [cited by examiner]
US 20210097551A1 · Tzur · 2021 [cited by applicant]
US 20240135323A1 · Kumar · 2024 [cited by examiner]
US 20250036925A1 · Liu · 2025 [cited by examiner]
US 20250317370A1 · Dave · 2025 [cited by examiner]
CA 3042926A1 · 2019 [cited by examiner]
CN 110134881A · 2019 [cited by examiner]
CN 110516611A · 2019 [cited by examiner]
WO WO2023113878A1 · 2023 [cited by examiner]
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., . . . & Polosukhin, I. (2017). Attention is all you need. Advances in neural information processing systems, 30. (Year: 2017). [cited by examiner]
Veličković, P., Cucurull, G., Casanova, A., Romero, A., Lio, P., & Bengio, Y. (2017). Graph attention networks. arXiv preprint arXiv: 1710.10903. (Year: 2017). [cited by examiner]
Cristian, M., Christian, S., & Dumitru-Tudor, T. (Sep. 2019). A study in the automation of service ticket recognition using natural language processing. In 2019 International Conference on Software, Telecommunications a… [cited by examiner]
Chen, Z., Liu, J., Su, Y., Zhang, H., Wen, X., Ling, X., . . . & Lyu, M. R. (Nov. 2021). Graph-based incident aggregation for large-scale online service systems. In 2021 36th IEEE/ACM International Conference on Automat… [cited by examiner]
Anonymous, “A Cognitive Method to Assist in Generating a Better Response in a Support System,” IP.com No. IPCOM000264315D, Dec. 2, 2020, 5 pages. [cited by applicant]
Han et al., “DeepRouting: A Deep Neural Network Approach for Ticket Routing in Expert Network,” 2020 IEEE International Conference on Services Computing, Beijing, China, Nov. 2020, 8 pages. [cited by applicant]
Mikolov et al., “Distributed Representations of Words and Phrases and their Compositionality,” Cornell University, Ithaca, NY, Oct. 2013, 9 pages. [cited by applicant]
Wahba et al., “Evaluating the Effectiveness of Static Word Embeddings on the Classification of IT Support Tickets,” CASCON '20: Proceedings of the 30th Annual International Conference on Computer Science and Software En… [cited by applicant]
Bansal et al., “IT Ticket Classification,” Analytics Insight, Aug. 22, 2020, 15 pages. [cited by applicant]
Gupta et al., “Multi-dimensional Knowledge Integration for Efficient Incident Management in a Services Cloud,” 2009 IEEE International Conference on Services Computing, Bangalore, India, Sep. 2009, pp. 57-64. [cited by applicant]
Grover et al., “node2vec: Scalable Feature Learning for Networks,” Cornell University, Ithaca, NY, Jul. 2016, 10 pages. [cited by applicant]