Evaluating the Quality of Benchmark Datasets for Low-Resource Languages: A Case Study on Turkish
- URL: http://arxiv.org/abs/2504.09714v2
- Date: Sat, 26 Apr 2025 11:28:53 GMT
- Title: Evaluating the Quality of Benchmark Datasets for Low-Resource Languages: A Case Study on Turkish
- Authors: Ayşe Aysu Cengiz, Ahmet Kaan Sever, Elif Ecem Ümütlü, Naime Şeyma Erdem, Burak Aytan, Büşra Tufan, Abdullah Topraksoy, Esra Darıcı, Cagri Toraman,
- Abstract summary: This study addresses the need for robust and culturally appropriate benchmarks by evaluating the quality of 17 commonly used Turkish benchmark datasets.<n>Our results reveal that 70% of the benchmark datasets fail to meet our quality standards.<n>GPT-4o has stronger labeling capabilities for grammatical and technical tasks, while Llama3.3-70B excels at correctness and cultural knowledge evaluation.
- Score: 1.59623393716069
- License: http://creativecommons.org/licenses/by-nc-sa/4.0/
- Abstract: The reliance on translated or adapted datasets from English or multilingual resources introduces challenges regarding linguistic and cultural suitability. This study addresses the need for robust and culturally appropriate benchmarks by evaluating the quality of 17 commonly used Turkish benchmark datasets. Using a comprehensive framework that assesses six criteria, both human and LLM-judge annotators provide detailed evaluations to identify dataset strengths and shortcomings. Our results reveal that 70% of the benchmark datasets fail to meet our heuristic quality standards. The correctness of the usage of technical terms is the strongest criterion, but 85% of the criteria are not satisfied in the examined datasets. Although LLM judges demonstrate potential, they are less effective than human annotators, particularly in understanding cultural common sense knowledge and interpreting fluent, unambiguous text. GPT-4o has stronger labeling capabilities for grammatical and technical tasks, while Llama3.3-70B excels at correctness and cultural knowledge evaluation. Our findings emphasize the urgent need for more rigorous quality control in creating and adapting datasets for low-resource languages.
Related papers
- Do LLMs Understand Your Translations? Evaluating Paragraph-level MT with Question Answering [68.3400058037817]
We introduce TREQA (Translation Evaluation via Question-Answering), a framework that extrinsically evaluates translation quality.
We show that TREQA is competitive with and, in some cases, outperforms state-of-the-art neural and LLM-based metrics in ranking alternative paragraph-level translations.
arXiv Detail & Related papers (2025-04-10T09:24:54Z) - Synthetic Data Generation for Culturally Nuanced Commonsense Reasoning in Low-Resource Languages [5.376127198656944]
We compare three dataset creation strategies: (1) LLM-assisted dataset generation, (2) machine translation, and (3) human-written data by native speakers, to build a culturally nuanced story comprehension dataset.<n>Our findings indicate that LLM-assisted data creation outperforms machine translation.
arXiv Detail & Related papers (2025-02-18T15:14:58Z) - Multilingual European Language Models: Benchmarking Approaches and Challenges [2.413212225810367]
generative large language models (LLMs) can solve different tasks through chat interaction.<n>This paper analyses the benefits and limitations of current evaluation datasets, focusing on multilingual European benchmarks.<n>We discuss potential solutions to enhance translation quality and cultural biases, including human-in-the-loop verification and iterative translation ranking.
arXiv Detail & Related papers (2025-02-18T14:32:17Z) - Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual Evaluation [71.59208664920452]
Cultural biases in multilingual datasets pose significant challenges for their effectiveness as global benchmarks.<n>We show that progress on MMLU depends heavily on learning Western-centric concepts, with 28% of all questions requiring culturally sensitive knowledge.<n>We release Global MMLU, an improved MMLU with evaluation coverage across 42 languages.
arXiv Detail & Related papers (2024-12-04T13:27:09Z) - Correcting FLORES Evaluation Dataset for Four African Languages [2.552967468434151]
The original dataset, though groundbreaking in its coverage of low-resource languages, exhibited various inconsistencies and inaccuracies.
Through a meticulous review process by native speakers, several corrections were identified and implemented.
We believe that our corrections improve the linguistic accuracy and reliability of the data.
arXiv Detail & Related papers (2024-09-01T06:13:03Z) - The BiGGen Bench: A Principled Benchmark for Fine-grained Evaluation of Language Models with Language Models [94.31327813151208]
BiGGen Bench is a principled generation benchmark designed to thoroughly evaluate nine distinct capabilities of LMs across 77 diverse tasks.
A key feature of the BiGGen Bench is its use of instance-specific evaluation criteria, closely mirroring the nuanced discernment of human evaluation.
arXiv Detail & Related papers (2024-06-09T12:30:30Z) - QuRating: Selecting High-Quality Data for Training Language Models [64.83332850645074]
We introduce QuRating, a method for selecting pre-training data that can capture human intuitions about data quality.
In this paper, we investigate four qualities - writing style, required expertise, facts & trivia, and educational value.
We train a Qur model to learn scalar ratings from pairwise judgments, and use it to annotate a 260B training corpus with quality ratings for each of the four criteria.
arXiv Detail & Related papers (2024-02-15T06:36:07Z) - NusaWrites: Constructing High-Quality Corpora for Underrepresented and
Extremely Low-Resource Languages [54.808217147579036]
We conduct a case study on Indonesian local languages.
We compare the effectiveness of online scraping, human translation, and paragraph writing by native speakers in constructing datasets.
Our findings demonstrate that datasets generated through paragraph writing by native speakers exhibit superior quality in terms of lexical diversity and cultural content.
arXiv Detail & Related papers (2023-09-19T14:42:33Z) - SEAHORSE: A Multilingual, Multifaceted Dataset for Summarization
Evaluation [52.186343500576214]
We introduce SEAHORSE, a dataset for multilingual, multifaceted summarization evaluation.
SEAHORSE consists of 96K summaries with human ratings along 6 dimensions of text quality.
We show that metrics trained with SEAHORSE achieve strong performance on the out-of-domain meta-evaluation benchmarks TRUE and mFACE.
arXiv Detail & Related papers (2023-05-22T16:25:07Z) - Beyond Counting Datasets: A Survey of Multilingual Dataset Construction
and Necessary Resources [38.814057529254846]
We examine the characteristics of 156 publicly available NLP datasets.
We survey language-proficient NLP researchers and crowd workers per language.
We identify strategies for collecting high-quality multilingual data on the Mechanical Turk platform.
arXiv Detail & Related papers (2022-11-28T18:54:33Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.