Logo: to the web site of Uppsala University

uu.sePublications from Uppsala University
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
A Generative AI Chatbot Poweredby Gemini 1.5: Optimized andEvaluated for Real Estate Inquiriesin Sweden: An Examination of LLM System Optimization Techniquesfor Large-Context LLM Models
Uppsala University, Disciplinary Domain of Science and Technology, Mathematics and Computer Science, Department of Information Technology.
2024 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

This thesis explores the potential of large language models to create a cost-effective and practicalsupport chatbot for the real estate industry. The goal is to develop a solution that assists users byproviding relevant information tailored to industry standards and individual user data, deliveredthrough an engaging chat interface. The use of RAG (Retrieval-Augmented Generation) isinvestigated as a means to improve the accuracy and reliability of the chatbot by providingaccess to real-time updates and a diverse range of data sources, including legal documents,contracts, and publicly available online information. The effectiveness of the chatbot is evaluatedbased on response accuracy, latency, and cost-efficiency. The study makes use of metrics such asMean Average Precision, Normalized Discounted Cumulative Gain and Hit Rate to assess theperformance of the vector search solution used for RAG. We conclude that optimizationtechniques like RAG, system instruction- and prompt -engineering are vital to ensure highresponse quality, user experience and cost-efficiency.

Place, publisher, year, edition, pages
2024.
Series
IT ; mDV 24 024
Keywords [en]
llm, ai, python, nlp
National Category
Computer Sciences
Identifiers
URN: urn:nbn:se:uu:diva-539274OAI: oai:DiVA.org:uu-539274DiVA, id: diva2:1901253
External cooperation
Pigello AB
Supervisors
Examiners
Available from: 2024-09-26 Created: 2024-09-26 Last updated: 2024-09-26Bibliographically approved

Open Access in DiVA

No full text in DiVA

By organisation
Department of Information Technology
Computer Sciences

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 1505 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf