Ottili LD3

LD3 topic verification

Before a topic becomes a content angle, LD3 checks it for strategy alignment, duplicates, source trust and internal conflict. Weak, critical or duplicate topics are flagged, not silently used.

Overview

Topic verification is the gate between detected topics and the content angles LD3 later produces. Before a detected topic becomes a publishable angle, LD3 checks it across four dimensions: strategy alignment, duplicates, source trust and internal conflict. Topics that fail the check are flagged and surfaced to a reviewer — not silently fed into generation.

Where topic verification stands

Ottili LD3 is currently in public Beta*. The capabilities source intake, verification, facts, topic expansion, content packages, review, publishing, analytics and learning are available on the public LD3 page (/ld3). Topic verification runs as part of the topic and verification pipeline automatically, once sources are connected.

What topic verification checks

Strategy alignment* — LD3 compares the detected topic against your configured core topics, excluded topics and content goal (strategy_alignment_score). Topics outside the strategy are flagged instead of used blindly.
Duplicates* — LD3 looks for similar existing topics in the same project. At a similarity of 0.85 (threshold topic_deduplication_threshold) the topic is flagged as a duplicate; the action is block or merge (topic_deduplication_action).
Source trust* — the sources behind the topic receive trust, recency and certainty scores (_source_trust_score, _recency_score, _visual_certainty_score).
Internal conflict* — the topic's claims are matched against evidence. On conflicting evidence LD3 sets conflict_flag and lists conflicting_evidence_ids; the conflict_ratio is checked against maximum_conflict_ratio (default 0.5).

What happens to weak or critical topics

Topics with low trust, high conflict, detected duplicates or critical content (is_critical_topic) are flagged, not silently used*. Expansion only builds on topics you can stand behind; conflicts and duplicates are surfaced so a reviewer can decide how to handle them.

Why it matters

Topic verification protects the planning queue. When an angle is generated, it traces back to a topic that passed the check — and therefore to sources and facts that are attributable rather than merely plausible-sounding.

[LD3 topics](/docs/ld3-topics) — detecting and expanding topics into content angles.
[LD3 source verification](/docs/ld3-verification) — how LD3 scores individual sources.
[LD3 fact extraction](/docs/ld3-facts) — what happens to verified material next.
[LD3 reviews](/docs/ld3-reviews) — the human check before anything ships.

Was this article helpful?