Skip to main content
After the reasoning analysis, Fracture runs a separate source pass that searches the public web for your factual claims. This step is designed to surface candidates for verification — it tells you where support likely exists and where you still need to do your own confirming.

​
How It Works

1

Claim Extraction

The engine pulls out quoted passages, the claims it flagged during analysis, and factual-looking sentences from your draft.
2

Public Web Search

For each extracted claim, the engine searches public pages and retrieves page text to compare against what your draft asserts.
3

Match Scoring

The engine compares the retrieved page against your claim and assigns a support status based on how closely the page’s content aligns with your assertion.
4

Citation Building

For well-matched claims, the engine builds a starter MLA or APA entry from the page’s details, giving you a draft citation to refine.

​
Support Statuses

​
What a Match Measures

A match score reflects how closely your claim’s distinctive terms appear on the retrieved page. A high match means the page discusses the same subject in similar terms — not that the page proves your specific claim. The engine evaluates textual overlap, not source authority or methodological credibility.
Every match is tentative. Open the linked page and confirm the exact passage before using it in a final paper or speech.