🔍 What Is a Semantic Triple?
A semantic triple is a structured way of expressing a single fact by linking three elements together. Scholar's factuality grader reads your content and attempts to extract these triples automatically. The more triples Scholar can detect and confirm, the higher your factuality score.
There are two common formats Scholar recognises:
- SPO (Subject–Predicate–Object): The most widely used format. Example — Albert Einstein (Subject) developed (Predicate) the theory of relativity (Object).
- EAV (Entity–Attribute–Value): A variation often used to describe properties. Example — Mount Everest (Entity) height (Attribute) 8,849 metres (Value). EAV triples are especially common in content that lists product specs, statistics, or scientific measurements.
Both formats serve the same purpose: they let Scholar isolate discrete, checkable claims inside your content rather than analysing vague or subjective language.
📐 How Scholar Calculates Your Factuality Score
When you run a Scholar report, the grader scans your content and attempts to do the following:
- Extract triples — Scholar parses your text and identifies statements that fit an SPO or EAV pattern.
- Verify triples — Each extracted triple is checked to confirm the claim is factually accurate.
- Score your content — Your factuality score reflects how many of the detected triples can be confirmed as accurate. A higher proportion of confirmed triples produces a higher score.
The triple count shown in your report reflects how many triple-shaped statements Scholar detected in your content. This number alone does not determine your score — what matters is whether those triples can be verified.
⚠️ Why Your Factuality Score Shows as 0
A factuality score of 0 does not mean your content contains no facts. It means Scholar was unable to verify any of the detected triples. The most common reasons this happens are:
- No verifiable triples detected: Scholar found statements that look like triples in structure but could not isolate complete SPO or EAV patterns — often because the language is too vague, opinion-based, or heavily qualified.
- Triples could not be confirmed: Scholar extracted triples successfully but the specific claims (dates, figures, names, relationships) could not be verified. This can happen with highly niche, proprietary, or newly published information.
- Content is non-factual in nature: Purely promotional, narrative, or opinion-driven content typically contains few or no verifiable factual claims, resulting in a score of 0.
✅ What You Can Do
If your factuality score is unexpectedly 0, try the following:
- Review your content for concrete, specific claims — named entities, measurable figures, or established facts tend to produce extractable triples.
- Reduce heavy qualification or hedging language that may prevent Scholar from isolating clear SPO or EAV patterns.
- Re-run the Scholar report after making revisions to see whether the score updates.
- If the score remains 0 despite your content containing clear factual statements, this may indicate a grading issue that requires investigation by our team.
If you need further assistance, open the chat widget in the bottom-right corner of the platform and type human teammate to be connected with a member of our team.