Troubleshooting bot, machine learning, deep learning
Currently, many IT companies are providing technical support either for their own products, or for general technical demands of customers. The troubleshooting is sort of technical support, which involve human interaction to handle such issues. When the product is owned by the company, support engineers have to take extensive training in order to understand the concepts and the general technical issues. Nonetheless, the training cannot cover all possible cases and they have to learn by experience or by senior colleagues.
Taking into consideration that supports teams of a company use some ticketing system that allows archiving history and they most probably have historical records, which can be transformed into the knowledge base. In this thesis work, we will address the challenge of developing an automated troubleshooting bot, which is capable of taking advantage of a large amount of data that consists of previously arisen problems and their solutions. We will introduce the data engineering layer which transforms the support historical records into informative data. In addition, we will leverage machine learning algorithms for clustering and find similarities within the former solutions. Finally, we will work on test understanding and paraphrasing to give more compressive answers to assist the support engineers in their job.
- python, machine learning, statistics, data analysis,image data
- Artificial Intelligence
- Heterogeneous Dataset, opencv, deep learning, Autoencoders, GANs
gépi tanulás, machine learning, deep learning, GANs