Original synthetic source note
The imaginary Sample Desk runs on macOS and stores results locally. Linux support is the next development task.
Knowledge & local AI
FactShort
Follow a fact-check from the claim to the evidence and assessment.
Interactive study
Explore the exampleWhy I’m interested
A fact-check becomes useful when you can follow the reason for its verdict. I want the claim, cited evidence, and assessment in the same view, so you can inspect the connection yourself. Evidence that contradicts a claim and evidence that says nothing about it need different answers.
Progress
The Python application includes a web desk, CLI, adapters, and a shared claim-research pipeline. The Rust page displays prewritten assessments of a synthetic note: supported, contradicted, and no evidence. The next integration brings an offline report from the Python pipeline into this view.
Three claims, one source
Put three claims against the same synthetic note. Follow the evidence to see why each gets a different assessment.
A claim stays beside its evidence
The source says the application stores results locally, so it supports the claim.
Input
The Sample Desk stores results locally.Result
Assessment: Supported
Reason: The source says the application stores results locally, so it supports the claim.Source: Read the matching passage
A claim stays beside its evidence
The source lists Linux support as the next development task, which contradicts the claim that it is complete.
Input
Linux support is complete.Result
Assessment: Contradicted
Reason: The source lists Linux support as the next development task, which contradicts the claim that it is complete.Source: Read the matching passage
A claim stays beside its evidence
The source gives no user count. The assessment records the missing evidence.
Input
The Sample Desk has ten thousand users.Result
Assessment: Insufficient evidence
Reason: The source gives no user count. The assessment records the missing evidence.Source: Read the matching passage
The assessments are prewritten for this synthetic source. Rust exports the report views. FactShort's Python pipeline supplies the claim-research application.
Sources and reference notes
What comes next
Export an offline report from the Python pipeline and display its claims, sources, and assessments through the same report view.
Implementation and credits
What evidence supports, contradicts, or leaves a claim unanswered?
Prewritten assessments evaluate three claims against one original synthetic source note.
Compare the supported claim, contradiction, and missing evidence, with the source beside each assessment.
Rust exports the report example. Python implements the claim-research pipeline.
A FactShort report example with an original synthetic source note.