Structured Knowledge Publishing

Make company information easier for AI to read.

Add sources to the company information you publish, then produce web pages and structured data from the same content. People and AI can read and check the same facts.

FactMind

One source, multiple reading formats

What is knowledge publishing

Give AI sourced information to use when describing your company.

Organize company, product and policy information and publish it with supporting sources. The web page and its machine-readable data describe the same facts.

Sources: STRUCTURED-DATA · AI-FEATURES

  1. Official sourceFacts, sources, publication scope
  2. Publish formatsHTML and structured data
  3. AI readsConsistent meaning and sources
  4. UpdateRefresh outputs and corrections

Why it matters

Information to discover

Publish public pages and paths such as sitemaps so search engines and AI services can find company information.

Meaning to read

State the relationships among company names, products, policies and sources in structured data so AI and search engines have cues to interpret them.

Facts that stay aligned

Carry the same correction into the explanations and data.

The difference sources can make

Adding sources can change how much an answer draws from a page.

GEO compared original content with a version augmented by source citations.

1,000 GEO-bench test queries · top five Google documents · GPT-3.5-turbo · temperature 0.7 · five responses.

How adding sources affects visibility in generated answers The visibility score rose from 19.3 to 24.6, a 5.3-point increase.

The visibility score rose from 19.3 to 24.6, a 5.3-point increase.

Position-Adjusted Word Count — Overall weights cited answer text by length and position. It is not a click rate or citation probability.

View the comparison data
How adding sources affects visibility in generated answers
ConfigurationSource visibility score
No Optimization · original content19.3
Cite Sources · citations added24.6

Aggarwal et al. · GEO · KDD 2024 / arXiv v3 · Table 1

Values are from v3 Table 1; separate appendix values and deviations are not combined here.

External content-editing research; not a FactMind or JSON-LD measurement, or a guarantee of AI citations.

FactMind establishes a consistent source of company information.

BLUEURBAN studies how to show the sources of company information and keep pages consistent with structured data. FactMind creates several formats from one approved source and updates them together.

Research behind the technology

Research and application at BLUEURBAN

FactMind: prepare company facts once and publish them in several formats.

FactMind turns company-approved information into HTML, JSON-LD, facts.json and answer pages. It verifies control of the site before publishing and updates the formats and correction records together when information changes.

Illustrative example

When company information changes, its published forms should agree.

An official update

The company changes its published contact address.

FactMind

  • Check the approved source and control of the site.
  • Update each format and the correction record from that source.

Publications from one source

HTML
A readable contact page
JSON-LD
Contact information for the page
facts.json
Facts and sources in FactMind’s format

Information AI can read

  • Read the explanation on the web page.
  • Find the same fact and source in structured data.
Confirm what to publish and its sources
Approve the publication scope and record the source and responsible party for each fact. Verifying site control is not the same as judging whether a fact is true.
Prepare once, publish in several forms
Create page explanations and data fields from the same information. facts.json is a FactMind format, not a universal AI standard.
Update all formats when facts change
Update related outputs together and use sitemaps and IndexNow to communicate page locations and changes.
BLUEURBAN engineering blog
Applied at BLUEURBANFactMindFacts and sourcesPublic webHTML · JSON-LD · facts.json

Frequently asked questions

Q1

Will AI necessarily cite a published page?

No. Each AI service decides what it collects and cites. FactMind publishes sourced facts in forms that are easy to read and inspect; it cannot control citation decisions.

Q2

Is adding JSON-LD enough?

Visible content and structured data need to agree, and pages need to be discoverable. Sources and updates also matter.

Q3

Does approval certify truth?

Approval records publication scope and responsibility. Site control does not certify content accuracy.

Papers and resources

  1. GEO: Generative Engine Optimization

    KDD 2024 · arXiv v3, 2024-06-28

  2. Introduction to structured data markup in Google Search

    Google Search Central

  3. AI features and your website

    Google Search Central

  4. IndexNow FAQ

    IndexNow official documentation

Publish company information consistently.

Review what to publish, who maintains it, its formats and its correction path.

Discuss knowledge publishing