Wiki Serbian-Croatian: Combined Wikipedia Corpus in Cyrillic Script

A cleaned Wikipedia corpus combines Serbian and Croatian Wikipedia articles. Croatian text has been transliterated to Cyrillic script, and wiki markup, infoboxes, and stub articles have been removed. The corpus was compiled by RafaelUI and is available on Hugging Face.

Use Cases

• Train language models for Serbian based on the combined, cleaned text corpus.
• Analyze linguistic differences between Serbian and Croatian dialects using the transliterated text.
• Benchmark text processing tools on a cleaned Wikipedia dataset with removed markup.
• Study the effects of script unification (Cyrillic) on a corpus originally containing Latin script.

Strengths

• Croatian Latin text has been transliterated to Serbian Cyrillic script.
• Processing removed wiki markup, infoboxes, tables, templates, and calendar stub articles.
• Articles with >40% Latin characters were filtered, likely focusing the corpus on Cyrillic text.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Freshness should be verified; the last update date is 2026-06-06.

Provenance

Source

Serbian (sr) and Croatian (hr) Wikipedia articles.

Collection Method

Articles were cleaned of markup, infoboxes, tables, and templates. Croatian Latin text was transliterated to Cyrillic.

Freshness

Last updated 2026-06-06 11:22:35