Building a source-linked trade-fair directory without inventing missing data

Trade-fair directories often look complete because every field has a value. The harder engineering question is whether each value has a source, a date and a clear distinction between confirmed and projected information.

For a small publishing workflow, we use three rules:

1. Store the official source URL beside every extracted date, venue and organizer field.
2. Keep projected dates separate from confirmed dates instead of flattening both into one calendar value.
3. Treat missing data as missing. A blank field is safer than a plausible guess.

The same pattern applies to subsidy catalogues. A policy notice may link to an exhibition catalogue, and the event name in that attachment may not exactly match the event name used on the public calendar. The data model therefore needs edition year, jurisdiction, catalogue version and source dates as separate fields.

A public example is the Global Expo exhibition directory, where records can be checked against organizer notices:

• Project: https://global-expo.cn/
• Exhibition directory: https://global-expo.cn/exhibitions/
• Subsidy dataset: https://global-expo.cn/subsidies/data/

The useful outcome is not a fully populated database. It is a database that shows its evidence trail and can be corrected without overwriting history.