التفاصيل البيبلوغرافية
| العنوان: |
Autodispatcher: a fully automated failure analyzer |
| المؤلفون: |
Vieira, Daniela Domingos |
| المساهمون: |
Garcia, Nuno Ricardo da Cruz, Guimarães, Luís Manuel Rodrigues, Repositório da Universidade de Lisboa |
| سنة النشر: |
2024 |
| مصطلحات موضوعية: |
Machine Learning, Redes Neuronais, Árvores de Decisão, Python, TensorFlow, Trabalhos de projeto de mestrado - 2024, Departamento de Informática |
| الوصف: |
Trabalho de projeto de mestrado, Ciência de Dados , 2023, Universidade de Lisboa, Faculdade de Ciências |
| Description (Translated): |
In order to minimize the impact of potential failures of assets within its telecommunications infrastructure in Portugal, NOS is interested in improving the efficiency of current decision processes which are led by human operators. Currently, an expert team manually selects response teams to deal with occurred failures, based on a pre-compiled set of policies to classify alarms. However, NOS aims to leverage ML-based techniques to help the expert team select the most appropriate response team to deal with an occurred technical issue. Towards that goal, one of the main challenges lies in the highly dynamic nature of the alarms issued by the company’s large and diverse infrastructure. In this project, we explore the design of a fully automated failure analyzer based on neural networks, called Autodispatcher, to tackle this challenge. The Autodispatcher framework will function as a decision tree, capable of efficiently analyzing and processing real-time alarms emitted by the infrastructure monitoring tools. It is designed to have four stages, including two stages which are the main focus of this work: a string frequency analysis method and a neural network approach for word analysis. To ensure the efficiency and effectiveness of the Autodispatcher, regular model updates and retraining are conducted. This allows the system to adapt to changes in the infrastructure and incorporate new data patterns that may arise over time. Additionally, human operators play a considerable role in supervising and validating the Autodispatcher’s decisions, because they have the ability to review and adjust the system’s recommendations. The main goal of the Autodispatcher is to minimize response time and optimize resource allocation. By automating the process of selecting response teams, NOS can significantly improve the efficiency of its operations. This reduces the impact of failures on the infrastructure and enhances the overall customer experience by ensuring timely resolutions to technical issues. |
| وصف الملف: |
application/pdf |
| اللغة: |
English |
| الاتاحة: |
http://hdl.handle.net/10451/64342 |
| Rights: |
open access |
| رقم الانضمام: |
rcaap.com.ul.repositorio.ul.pt.10451.64342 |
| قاعدة البيانات: |
RCAAP |