Summary
Asia-Pacific Network Operations and Management Symposium
2022
Session Number:PS3
Session:
Number:PS3-17
An NLP-Based Model for the Dispatching Problem of Network Operation
Li Yun-Jie, Li Jhao-Yin,
pp.-
Publication Date:2022/09/28
Online ISSN:2188-5079
PDF download
Summary:
It is difficult for Telecom to decide whether dispatch the engineer to fix the network problem when the network management system can not figure out the detail of the problem. Some network problems may not derive from the network devices because those problems may trigger by other factors (e.g., an external power source failure, weather issue). However, the user would report that they can not use the network service and Telecom have to decide whether ask the engineer to go out to deal with the problem. If the engineer goes out to handle this problem but figures out there is nothing to do, it would waste their time and it means that this wrong decision would increase the Operating Expenses (OPEX).In this study, we would like to use the text data that the users report to Telecom to try to describe the failure from their perspective. The purpose is that we would like to take the data as a Natural Language Processing problem (NLP) by using BERT model to check that is how it can improve this problem. Though the result showing the model can get better performance seems to not significant improvement on this problem, we conduct the preliminary experiment on the real-world data to show that it is feasible of using only the text data to help Telecom to reduce the cost of network operation.