1. Ask
You write a question in your own words, in any of 280+ languages, or pick a topic.
How it works
Four steps, no black box. Every sentence you read was written by a Wikipedia editor and is quoted, not generated.
2. Research
indusmova queries the Wikipedia search API, resolves your question to the article that fits best, and pulls the full article and its metadata.
3. Filter
The article is cleaned — navigation boxes, edit links and footnote markers dropped — then split into a lead and topic sections, and its sentences are scored against your question.
4. Understand
You get a direct answer, key points, a topic-wise breakdown, the infobox as facts, and every source with a link back to Wikipedia.
Which APIs are used
- Wikipedia Action API — search, article parsing, sections, random articles, prefix suggestions, plain-text extracts.
- Wikipedia REST API — page summaries, related articles, the On this day feed, featured content.
- Wikimedia pageviews API — the "most read right now" list.
- Wikidata — structured facts behind a topic.
- Wikimedia Commons — image credits and licences.
All of these are public, need no key, and are called directly from your browser. There is no indusmova server in the path.
How sentences are chosen
The default ranker is a keyword overlap score: each word in your question is weighted by how rare it is across the article, and sentences that match are scored higher. A sentence containing a date or a number, and one that opens a paragraph, get a small bonus. The optional Understand meaning switch instead loads a small open sentence-embedding model (about 23 MB, once) and re-orders the same sentences by cosine similarity to your question. Either way, the sentences themselves are never rewritten.
What indusmova never does
- It does not generate prose, summaries or claims of its own.
- It does not store your questions on a server — there is no server.
- It does not track you, sell data, or show advertising.
- It does not claim to be affiliated with the Wikimedia Foundation.
Licences
Wikipedia article text is available under the Creative Commons Attribution-ShareAlike Licence (CC BY-SA 4.0). Wikidata data is released under CC0 1.0. Files on Wikimedia Commons carry their own licence, shown with each image. Reuse must respect those terms.