Comparing AI models in graph database development

Mohsen Deeb

Bakalářská práce

Abstract:

This study compares the performance of ChatGPT and Gemini in graph database development through a structured evaluation process. A dataset of 332 publications was retrieved from Scopus, and citation relationships were analyzed using VOSviewer. Both AI models were tasked with generating, optimizing, and executing graph queries to identify key authors and citation patterns. Their performance was assessed based on key metrics such as query accuracy, execution time, query simplicity, and optimization effectiveness. The results indicate that ChatGPT excels in generating syntactically correct and structured queries with high accuracy, making it more suitable for straightforward query tasks. However, Gemini demonstrates superior optimization capabilities, particularly for complex graph queries, reducing redundancy and improving execution efficiency. Statistical analysis confirms that Gemini produces more efficient query execution plans, especially in handling large-scale datasets and multi-hop relationships. This research provides valuable insights into the strengths and limitations of AI-assisted graph database development, offering practical recommendations for developers and researchers in selecting suitable AI models based on specific database tasks. The findings contribute to the broader field of AI-driven database optimization and highlight the potential of LLMs in advancing graph data analytics.

Keywords

Graph Databases, AI Models, Query Optimization

Jazyk práce: angličtina

Datum vytvoření / odevzdání či podání práce: 26. 3. 2025

Obhajoba závěrečné práce

• Vedoucí: doc. Ing. Vojtěch Merunka, Ph.D.
• Oponent: Martin Molhanec, Bc., externi

Česká zemědělská univerzita v Praze

Provozně ekonomická fakulta

Bakalářský studijní program:
Informatics