Preprocessing intent classification using an adaptive scripted chatbot for a changing cloud platform
Keywords
Loading...
Authors
Issue Date
2023-11-01
Language
en
Document type
Journal Title
Journal ISSN
Volume Title
Publisher
Title
ISSN
Volume
Issue
Startpage
Endpage
DOI
Abstract
Chatbots are being used more and more in our daily lives. For instance, chatbots are being used
to help with internal company processes, in addition to customer support and other areas. In this
research project, a chatbot is developed to help them navigate through a cloud platform. However,
this platform is changing in its structure, so the chatbot needs to adapt. Furthermore, a preprocessing
intent classification is needed to adapt to the changes of the cloud platform. In the related
work, no adaptive scripted chatbot with preprocessing intent classification was found. Therefore,
own possible versions of such a chatbot have to be developed. The versions are first discussed
theoretically before the most suitable version is implemented. The discussed versions include a
mixture of an open and a closed-domain chatbot as well as a pure closed-domain chatbot. Using
the Rasa framework, a scripted chatbot is developed that is adaptive and able to perform preprocessing
intent classification. This chatbot is a proof-of-concept version with satisfying results for
intent classification with a F1-score of 0.909, a precision score of 0.927 and an accuracy of 0.913
on the test dataset and for different test scenarios. This is a first step to have an adaptive scripted
chatbot with preprocessing intent classification, which can further developed for a large user group.
Keywords: chatbot, preprocessing, adaptiveness, open-domain chatbot, closed-domain chatbot
Description
Citation
Faculty
Faculteit der Sociale Wetenschappen