analiza aiqbench
This study exposes the illusion of data abundance for under-resourced languages like Lombard, revealing that web-scraped corpora are often plagued by misidentification, boilerplate noise, and severe dialectal bias. The findings underscore a critical shift in NLP development from quantity-driven scraping toward variety-aware, community-driven curation to ensure equitable model performance across diverse linguistic varieties.
"Chi nas dal soch el sent de legn" -- Auditing Text Corpora for Lombard
arXiv:2606.06349v1 Announce Type: new Abstract: Several of the world's languages are still under-resourced in terms of Natural Language Processing (NLP) tools. This is mostly due to the lack of high-quality datasets to train, develop, and evaluate systems and models for several tasks, such as Machine Translation (MT). We conduct a manual audit of the parallel and monolingual corpora available for Lombard, an under-resourced language continuum from Italy. Our analysis reveals that the perceived abundance of web-scraped data is an illusion, with massive datasets plagued by severe language misidentification, boilerplate text, and non-linguistic noise. Furthermore, we analyze the orthographic composition of the valid Lombard portions across web-scraped datasets, curated corpora, and benchmarks. Our findings show conflicting orthographical systems and severe representational bias across all corpora: high-quality data is heavily skewed towards Western Lombard varieties, with Eastern ones left on the margins. This underscores the need for variety-aware, community-driven data curation rather than purely quantity-driven scraping.
Abstract:Several of the world's languages are still under-resourced in terms of Natural Language Processing (NLP) tools. This is mostly due to the lack of high-quality datasets to train, develop, and evaluate systems and models for several tasks, such as Machine Translation (MT). We conduct a manual audit of the parallel and monolingual corpora available for Lombard, an under-resourced language continuum from Italy. Our analysis reveals that the perceived abundance of web-scraped data is an illusion, with massive datasets plagued by severe language misidentification, boilerplate text, and non-linguistic noise. Furthermore, we analyze the orthographic composition of the valid Lombard portions across web-scraped datasets, curated corpora, and benchmarks. Our findings show conflicting orthographical systems and severe representational bias across all corpora: high-quality data is heavily skewed towards Western Lombard varieties, with Eastern ones left on the margins. This underscores the need for variety-aware, community-driven data curation rather than purely quantity-driven scraping.
Computation and Language (cs.CL)
arXiv:2606.06349 [cs.CL]
(or arXiv:2606.06349v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2606.06349 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 244 cuvinte
Subiecte conexe
Auditing Chinese Web-scale Corpora via Sampled BPE Token Statistics
arXiv · 1 sept. 2026
Scores That Hold, Benchmarks That Leak: Measuring Dataset Contamination in Public Brain-Tumor MRI Classification
arXiv · 2 oct. 2026
From Association to Decision Value: Testing Paired Instruction Dependence for VLA Corpus Selection
arXiv · 22 oct. 2026
When a Data Artifact Isn't a Shortcut: Causal Auditing of Synthetic RLVR Corpora
arXiv · 2 oct. 2026