WikiFuse
Merges Wikipedia articles from multiple language editions into one comprehensive page, with sources preserved.
🔗 Visit WikiFuse
Description
Wikipedia isn't one encyclopedia — it's hundreds, and the French, Japanese, or Arabic article on a topic often contains facts, photos, or context the English one is missing (and vice versa). WikiFuse automates the tedious work of reading five language editions and manually merging what's useful: point it at a topic and it pulls every language version, translates the non-English ones, and intelligently combines overlapping content into a single, well-sourced page.
Under the hood it's a Python 3.11+ command-line tool that fetches articles via Wikidata QIDs, uses sentence-embedding similarity to align and deduplicate overlapping claims across languages, and preserves reference attribution back to the original source article. It outputs MediaWiki wikitext, HTML, or an intermediate JSON for further processing, plus a side-by-side comparison view for reviewing what came from where. An optional LLM-enhanced merge mode (via your own OpenAI key) improves fusion quality at the cost of API spend.
💬 Our review
The short version: a genuinely useful niche tool for anyone who edits Wikipedia or researches a topic across languages — it automates a real, tedious task, though its usefulness depends entirely on whether translation quality holds up on your topic.
There's no direct competitor doing exactly this — general documentation tools like Pandoc or MkDocs convert formats but don't merge multilingual sources with semantic deduplication. WikiFuse's honest limitation is that its output quality rides on machine translation and embedding-based alignment, both of which can introduce subtle errors on nuanced or contested topics — so treat the merged page as a strong first draft, not a publish-ready final version. Free and open source (MIT), with LLM-enhanced mode as the only real cost if you opt in. Worth it for Wikipedia editors and multilingual researchers; overkill for casual use.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Gratuit (MIT) ; coûts API OpenAI optionnels pour le mode LLM.
Pros
Fusion multilingue automatisée
Alignement sémantique par embeddings
Attribution des sources préservée
Cons
Qualité dépend de la traduction
Nécessite un QID Wikidata
Mode LLM payant en option