Abstract
This master's thesis serves as a case study for the development of AI assistance for complex,configurable, and diverse product portfolios, using the example of searching for immediately
available new cars based on textual equipment descriptions. The data foundation consists of data
from a QA system. In combination with the methods of the current state of science and
technology, at least one prototype is developed based on these data, considering costs and
response times.
The development of an evaluation for this specific application is also an essential component of
this work and is intended to provide insights into the quality achieved in the developed prototype.
The prototype was developed with a focus on the RAG (Retrieval Augmented Generation)
system, based on the reasoning that it offers a promising starting point for potential further
developments. The results of this work demonstrate that the RAG system's schema is well-suited
for implementation in applications of this kind, with the Large Language Model being used for a
single response only.
| Date of Award | 2024 |
|---|---|
| Original language | German (Austria) |
| Supervisor | Ulrich Bodenhofer (Supervisor) |