Licensed, structured, ML-labeled Russian news from 105+ sources. Urgency scoring, NER, sentiment, event clustering — ready for LLM training, RAG, and financial risk models.
Not just raw text. Every article is structured, scored, clustered, and labeled before it hits the API.
cluster_size tells you how many outlets covered it — a direct signal strength indicator.[{text, type: PER/ORG/LOC}] — plug directly into any NLP pipeline.A single API call returns everything your model needs — no secondary enrichment required.
| Field | Type | Description |
|---|---|---|
| id | int | Unique article ID |
| title | str | Headline (Russian) |
| source | str | Publisher name (ТАСС, РИА, BBC, NEXTA…) |
| source_type | str | rss / telegram — ingestion method |
| fetched_at | datetime | Ingestion timestamp (UTC) |
| published_at | datetime | Original publication time |
| urgency_score | float 0–10 | ML urgency score |
| urgency_label | str | routine / elevated / breaking / urgent |
| cluster_id | int | Event cluster — same story grouped across sources |
| cluster_size | int | Coverage count — how many sources reported this |
| category | str | politics / conflict / economy / disaster / society… |
| geo_region | str | russia / ukraine / belarus / intl / other |
| language | str | ru / en / uk (auto-detected) |
| topic_tags | str[] | Top-3 topic tags: war / sanctions / energy / elections… |
| link | str | Original article URL |
| content_preview | str | First 200 chars (Standard+ gets full text) |
| sentiment | str Pro | positive / negative / neutral (ML classifier) |
| entities | obj[] Pro | NER: [{text, type: PER/ORG/LOC, freq}] |
| keywords | str[] Pro | Top-10 TF-IDF keywords from article text |
| summary_en | str Enterprise | AI-generated English summary (3 sentences) |
| translation_en | str Enterprise | Full English translation |
| parquet_url | str Enterprise | Parquet file URL for batch ML pipelines |
Both state-controlled and independent media. RSS feeds + live Telegram channel scraping.
All tiers include API access, JSONL export, and urgency scoring. Upgrade anytime via Telegram.
Request a free 1,000-article JSONL sample with full schema. No commitment. Cancel anytime.