Private digital intelligence bureau Lawful OSINT · Cyber monitoring · Due diligence

Public research resources · FR / EN

Public data and statistics · OSINT

Can these figures be compared without changing their meaning ? Practical method, external tools, source log and interpretation limits.

External services · no results hosted here · queries sent only when you click a provider link.

Catalogue version · Published

Printed charts and public-data exploration

Can these figures be compared without changing their meaning ?

  1. Locate the publisher, original series and methodological documentation.
  2. Check unit, population, territory, period and revision date before comparing.
  3. Distinguish missing values, zero, estimates, series breaks and provisional data.
  4. Retain series code, filters, version and licence with the table used.

Keep these fields in the source log

  • Series code and publisher
  • Unit, population and period
  • Version, filters and licence

What the method cannot establish alone

An aggregate describes a population, not a person or transaction. An apparent increase can result from changed scope or methods.

External tools for this pathway

Entries draw on the providers’ portals or documentation. Each entry shows its latest review date. Check the provider for current access conditions, pricing and quotas.

Public data and statistics

data.gouv.fr

France · Public entry point

Find datasets, publishers, resources and reuse licences.

Limit: Publication on the portal does not replace publisher, method and date checks.

Record reviewed

Public data and statistics

data.europa.eu

European Union · Public entry point

Explore European catalogues and identify a dataset publisher.

Limit: Catalogues aggregate resources with varying quality and freshness.

Record reviewed

Public data and statistics

Eurostat

European Union · Public entry point

Compare European statistics using definitions, series codes and metadata.

Fields and checks to retain

Keep the dataset code, unit, geographical filters, period and table flags. Check provisional or estimated observations and any breaks in the series before calculating a trend.

Limit: Compare only compatible units, populations and periods.

Publisher reference · Record reviewed

Public data and statistics

World Bank Open Data

International · Public entry point

Find indicators by country, year and series code.

Limit: Distinguish estimates and revisions from observed measurements.

Record reviewed

Public data and statistics

OECD Data Explorer

International · Public entry point

Explore comparative series and their dimensions in OECD data.

Limit: Country coverage and methods vary by series.

Record reviewed

Public data and statistics

UN Comtrade

International · Access / quotas vary

Compare trade flows by reporter, partner, product and period.

Limit: Do not confuse trade statistics with proof of an individual transaction.

Record reviewed

Public data and statistics

CISA · Known Exploited Vulnerabilities

United States · Public entry point

Identify a CVE with documented real-world exploitation in the KEV catalog.

Limit: Absence from the catalog does not prove a CVE is not exploited. US federal deadlines are not universal deadlines.

Record reviewed

Public data and statistics

FIRST · EPSS

International · Public entry point

Consult a CVE exploitation estimate for the next thirty days.

Limit: The score measures neither business impact nor the compromise probability of your own system.

Record reviewed

Public data and statistics

Légifrance

France · Public entry point

Find a legal text and the version relevant to a date.

Fields and checks to retain

Record text identifier, article, version, displayed effective date and permanent link. Separate original and consolidated text.

Limit: Finding an article does not determine its application to a case; obtain competent analysis.

Record reviewed

Read the original record

Separate dataset, distribution and observation period

DCAT distinguishes a dataset from its distributions, such as a file or access service. Describe the resource actually consulted.

Fields to retain and interpretation limits
ItemRecord to retainLimit
DatasetIdentifier, publisher and temporal or geographic coverage.The update date is not the measured period.
DistributionVersion URL, format and access date.Two formats can expose different versions.
ReuseDistribution licence and applicable conditions.Public access does not mean unconditional reuse.

W3C — DCAT 3

Reading exercise · fictional scenario

Do more requests mean more incidents?

A portal reports more assistance requests this year. Its coverage and referral mechanisms also changed.

Before opening the explanation, separate what is observed from what is inferred. Identify the missing check that prevents the broader conclusion.

Show the reasoning and next check

Unsupported shortcut

Incidents increased by the same proportion.

Bounded conclusion

Recorded requests increased. A population-wide incident trend does not follow directly from this indicator alone.

Check that could change it

Compare definitions, coverage, duplicates and observation periods; seek a series collected using a stable method.

This exercise illustrates an interpretation error. It describes no real person, organisation or client case.

Questions about tools and external searches

Where are search results displayed?

On the selected provider’s website, in a new tab. IIS does not retrieve, host or display investigation results. Directory filters only filter the editorial catalogue.

Is the typed query stored?

The launchpad builds links inside your browser, without sending the query to IIS or storing it in browser storage. Clicking a prepared link sends the query to the named provider, under its own rules.

Are all tools free and available in every language?

No. Access may require an account, subscription, software or quota. The catalogue is available in French and English; providers set their interface languages, coverage and conditions.

Does no result prove that a fact is false?

No. Index coverage, spelling, language, access limits and capture dates can explain a missing result. Record what was searched and what remains unknown.