Disclosure
Method & curation
How this brief is made
This brief reads what Chinese primary sources publish — ministry releases, regulator notices, statistical bulletins, in the original — and works out what it means for Europe’s economic position. It is produced by a monitoring pipeline with AI assistance and filtered through an editorial perspective. Neither part is hidden: the machine finds, translates and drafts; a human decides what is published. An edition goes out every two to four weeks, when the new material carries an argument.
Collection
A curated network of primary Chinese, EU and multilateral sources — ministries, statistical offices, regulators and official publications — is scanned daily, weekly or fortnightly, depending on the source. A small number of think-tank and press feeds is carried alongside them and labelled by source; Chinese state media are read for what the leadership wants said, never as the sole source for a fact. Every signal links to the primary document it rests on.
Screening & scoring
Each item is first scored by rule-based term matching for European economic relevance, likely impact and confidence, then passed through an AI relevance gate designed to keep only signals with a genuine China–Europe nexus. Scores are shown as bands: High, Medium or Watch. A band is a triage aid, not a measurement. It ranks attention; it does not establish importance. Items from Chinese-language official sources that the keyword screen misses are escalated to the relevance gate, so a State Council opinion is not lost for failing to mention Europe in its title.
Translation & analysis (AI-assisted)
Chinese-language material is machine-translated, then re-screened on the English text. The reading of each signal — the mechanism by which it reaches Europe, the stakes, the transmission chain, the key facts, the strongest counter-reading and what would overturn it — is drafted by a large language model against a fixed set of verified regulatory references, not from the model’s own memory. Passages marked like this are claims not yet tied to a primary source. Machine-translated items are labelled as such until an editor confirms them.
Editorial curation
The automated monitor is raw, screened material — machine output, not curated, and kept out of search engines. Editionsare different: every signal in an edition is selected, read and released by a human editor, who can rewrite any line of the analysis; the lead read that opens an edition is the editor’s own argument, whatever draft it started from. Each item names its provenance — Chinese primary source, Chinese state media, EU official source, multilateral body, or analysis and press — and shows the Chinese original title where there is one. A signal that appeared in one edition is not repeated in the next. The framing reflects an explicit editorial perspective on how energy, materials and industrial transformation shape Europe’s position. That perspective is a lens, not a neutral summary. Where the evidence cuts against it, the brief says so.
Limits
Automated screening misses things and surfaces noise. Machine translation can be imprecise on technical or legal wording. Categories, sector labels and the impact grid are assigned automatically and are approximate. Nothing here is investment, legal or tax advice. Where a fact matters, follow the primary-source link and check it.
Behind this brief
Europe–China Signal is built, written and edited by Niklas Bals at Compact Strategy. If your organisation needs this kind of analysis applied to a concrete decision, start a conversation ↗.