LLM Performance in Low-Resource Languages: Selecting an Optimal Model for Migrant Integration Support in Greek

The integration of Large Language Models (LLMs) in chatbot applications gains momentum. However, to successfully deploy such systems, the underlying capabilities of LLMs must be carefully considered, especially when dealing with low-resource languages and specialized fields. This paper presents the...

Full description

Saved in:
Bibliographic Details
Main Authors: Alexandros Tassios, Stergios Tegos, Christos Bouas, Konstantinos Manousaridis, Maria Papoutsoglou, Maria Kaltsa, Eleni Dimopoulou, Thanassis Mavropoulos, Stefanos Vrochidis, Georgios Meditskos
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Future Internet
Subjects:
Online Access:https://www.mdpi.com/1999-5903/17/6/235
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849432886211510272
author Alexandros Tassios
Stergios Tegos
Christos Bouas
Konstantinos Manousaridis
Maria Papoutsoglou
Maria Kaltsa
Eleni Dimopoulou
Thanassis Mavropoulos
Stefanos Vrochidis
Georgios Meditskos
author_facet Alexandros Tassios
Stergios Tegos
Christos Bouas
Konstantinos Manousaridis
Maria Papoutsoglou
Maria Kaltsa
Eleni Dimopoulou
Thanassis Mavropoulos
Stefanos Vrochidis
Georgios Meditskos
author_sort Alexandros Tassios
collection DOAJ
description The integration of Large Language Models (LLMs) in chatbot applications gains momentum. However, to successfully deploy such systems, the underlying capabilities of LLMs must be carefully considered, especially when dealing with low-resource languages and specialized fields. This paper presents the results of a comprehensive evaluation of several LLMs conducted in the context of a chatbot agent designed to assist migrants in their integration process. Our aim is to identify the optimal LLM that can effectively process and generate text in Greek and provide accurate information, addressing the specific needs of migrant populations. The design of the evaluation methodology leverages input from experts on social assistance initiatives, social impact and technological solutions, as well as from automated LLM self-evaluations. Given the linguistic challenges specific to the Greek language and the application domain, research findings indicate that Claude 3.7 Sonnet and Gemini 2.0 Flash demonstrate superior performance across all criteria, with Claude 3.7 Sonnet emerging as the leading candidate for the chatbot. Moreover, the results suggest that automated custom evaluations of LLMs can align with human assessments, offering a viable option for preliminary low-cost analysis to assist stakeholders in selecting the optimal LLM based on user and application domain requirements.
format Article
id doaj-art-b6ab3fece33d4d7cb0a76d2b526d870c
institution Kabale University
issn 1999-5903
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Future Internet
spelling doaj-art-b6ab3fece33d4d7cb0a76d2b526d870c2025-08-20T03:27:14ZengMDPI AGFuture Internet1999-59032025-05-0117623510.3390/fi17060235LLM Performance in Low-Resource Languages: Selecting an Optimal Model for Migrant Integration Support in GreekAlexandros Tassios0Stergios Tegos1Christos Bouas2Konstantinos Manousaridis3Maria Papoutsoglou4Maria Kaltsa5Eleni Dimopoulou6Thanassis Mavropoulos7Stefanos Vrochidis8Georgios Meditskos9School of Informatics, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreeceSchool of Informatics, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreeceSchool of Informatics, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreeceSchool of Informatics, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreeceSchool of Informatics, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreeceDepartment of Theoretical & Applied Linguistics, School of English, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreecePRAKSIS, 10432 Athens, GreeceInformation Technologies Institute, Center for Research and Technology Hellas, 57001 Thessaloniki, GreeceInformation Technologies Institute, Center for Research and Technology Hellas, 57001 Thessaloniki, GreeceSchool of Informatics, Aristotle University of Thessaloniki, 54124 Thessaloniki, GreeceThe integration of Large Language Models (LLMs) in chatbot applications gains momentum. However, to successfully deploy such systems, the underlying capabilities of LLMs must be carefully considered, especially when dealing with low-resource languages and specialized fields. This paper presents the results of a comprehensive evaluation of several LLMs conducted in the context of a chatbot agent designed to assist migrants in their integration process. Our aim is to identify the optimal LLM that can effectively process and generate text in Greek and provide accurate information, addressing the specific needs of migrant populations. The design of the evaluation methodology leverages input from experts on social assistance initiatives, social impact and technological solutions, as well as from automated LLM self-evaluations. Given the linguistic challenges specific to the Greek language and the application domain, research findings indicate that Claude 3.7 Sonnet and Gemini 2.0 Flash demonstrate superior performance across all criteria, with Claude 3.7 Sonnet emerging as the leading candidate for the chatbot. Moreover, the results suggest that automated custom evaluations of LLMs can align with human assessments, offering a viable option for preliminary low-cost analysis to assist stakeholders in selecting the optimal LLM based on user and application domain requirements.https://www.mdpi.com/1999-5903/17/6/235chatbotGreek languagelarge language modelsLLM evaluationmigrant integration
spellingShingle Alexandros Tassios
Stergios Tegos
Christos Bouas
Konstantinos Manousaridis
Maria Papoutsoglou
Maria Kaltsa
Eleni Dimopoulou
Thanassis Mavropoulos
Stefanos Vrochidis
Georgios Meditskos
LLM Performance in Low-Resource Languages: Selecting an Optimal Model for Migrant Integration Support in Greek
Future Internet
chatbot
Greek language
large language models
LLM evaluation
migrant integration
title LLM Performance in Low-Resource Languages: Selecting an Optimal Model for Migrant Integration Support in Greek
title_full LLM Performance in Low-Resource Languages: Selecting an Optimal Model for Migrant Integration Support in Greek
title_fullStr LLM Performance in Low-Resource Languages: Selecting an Optimal Model for Migrant Integration Support in Greek
title_full_unstemmed LLM Performance in Low-Resource Languages: Selecting an Optimal Model for Migrant Integration Support in Greek
title_short LLM Performance in Low-Resource Languages: Selecting an Optimal Model for Migrant Integration Support in Greek
title_sort llm performance in low resource languages selecting an optimal model for migrant integration support in greek
topic chatbot
Greek language
large language models
LLM evaluation
migrant integration
url https://www.mdpi.com/1999-5903/17/6/235
work_keys_str_mv AT alexandrostassios llmperformanceinlowresourcelanguagesselectinganoptimalmodelformigrantintegrationsupportingreek
AT stergiostegos llmperformanceinlowresourcelanguagesselectinganoptimalmodelformigrantintegrationsupportingreek
AT christosbouas llmperformanceinlowresourcelanguagesselectinganoptimalmodelformigrantintegrationsupportingreek
AT konstantinosmanousaridis llmperformanceinlowresourcelanguagesselectinganoptimalmodelformigrantintegrationsupportingreek
AT mariapapoutsoglou llmperformanceinlowresourcelanguagesselectinganoptimalmodelformigrantintegrationsupportingreek
AT mariakaltsa llmperformanceinlowresourcelanguagesselectinganoptimalmodelformigrantintegrationsupportingreek
AT elenidimopoulou llmperformanceinlowresourcelanguagesselectinganoptimalmodelformigrantintegrationsupportingreek
AT thanassismavropoulos llmperformanceinlowresourcelanguagesselectinganoptimalmodelformigrantintegrationsupportingreek
AT stefanosvrochidis llmperformanceinlowresourcelanguagesselectinganoptimalmodelformigrantintegrationsupportingreek
AT georgiosmeditskos llmperformanceinlowresourcelanguagesselectinganoptimalmodelformigrantintegrationsupportingreek