Spotting Contradictions Without AI

Contradiction detection is the feature people assume requires AI. It doesn't.

The core idea

A contradiction is two statements about the same subject that can't both hold. So detection has two halves: (1) decide which statements are about the same thing, and (2) check whether they conflict.

Subject matching

We extract salient keywords from each sentence and require a meaningful overlap before comparing. This prevents the classic false positive where "revenue increased" and "costs decreased" get flagged despite being about different things.

Conflict rules

Then we apply polarity rules: increase vs. decrease, profit vs. loss, approved vs. rejected, compliant vs. non-compliant — plus affirmation/negation pairs and numeric mismatches for the same metric phrase.

It's transparent, fast, and you can read exactly why each pair was flagged.