Masterarbeit
Development of an Interaction Logging Mechanism for Web Chatbots
Completion
2025/02
Research Area
Students
Mohammed Maaz Gheewale
Advisers
Verena Traubinger M.Sc.
Dr.-Ing. Sebastian Heil
Description
In the evolving landscape of digital interactions, chatbots have become essential tools for enhancing user experiences across sectors like customer service, healthcare, and education. These conversational agents facilitate efficient exchanges, enabling tasks from simple information retrieval to complex problem-solving. The interaction necessary in these conversations can be for one rather detailed with singular interaction with single elements, but also include whole interaction patterns. The user experience can also be influenced negatively by usability smells, which could be solved by simple refactoring.
User logs can be used to automatically find these interaction patterns and usability smells. Interaction logging mechanisms provide systematic approaches to capture, store, and analyse interaction data, although specific approaches for chatbots are currently not available. Thus, this thesis aims to develop an advanced interaction logging mechanism tailored specifically for chatbots. The solution of this thesis should be able to provide user logging data which can be used to identify interaction patterns or usability smells present in chatbots.
The objective of this thesis is the creation of a solution or the combination of existing approaches to solve the problem of logging the user interaction in web chatbots. This comprises the analysis of the state of the art of existing logging tools and frameworks for user interactions, meaningful benchmarks of their performance, and relevant literature for their adaption to and isolation of chatbot interactions. From this analysis, a combined and improved prototypical implementation should demonstrate the feasibility of the chatbot specific logging mechanism. A suitable evaluation has to compare this prototypical implementation with existing solutions according to benchmarks extracted through the stat of the art additionally to an analyse of the logs on their feasibility to recognize interaction patterns and possible usability smells.
