Institutional Repository of Vilnius University
Agentais grįstas darbo rinkos modeliavimas: Taisyklėmis pagrįstų ir kalbos modeliais paremtų metodų lyginamasis tyrimas /
Abstract
dc:descriptionAgent-based modeling (ABM) has been a significant tool for understanding labor market dynamics. Most commonly used in ABMs are traditional rule-based methods. Their drawback is in capturing the complexity of human behavior within social systems. Recent advancements in large language models (LLMs), such as GPT, offer a novel approach to modeling these dynamics by simulating human-like reasoning and decision-making. This research aims to compare the effectiveness of rule-based and LLM-based approaches in ABMs of labor markets. The study utilizes agent-based simulations where agents represent individuals and firms within a labor market. Two models are developed: one using predefined rules for decision-making, and the other integrating LLMs to enhance adaptability and realism. The data for these simulations are derived from business-oriented social networks and statistical sources. The evaluation focuses on key labor market outcomes, including unemployment trends, job turnover rates, company profits. Results indicate that LLM-based agents perform worse compared to rule-based agents. Their performance strongly depends on the accuracy of the LLM used. Usage of LLMs requires complex answer parsing and error handling, since LLMs do not always provide answers in the requested format.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Stonkus, Arnas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD210648046&prefLang=en_US