` items have anchor tags leading to sources like “InfoSeemedia”, “Business Insider”, “People”. That’s good.
5. **GATE 4 — FACT LOCK VS VERIFIED_FACTS:** Since the provided `Verified facts (JSON)` is empty, there is nothing to override. I will skip this.
6. **GATE 5 — FAQ DEDUP:** The raw article has FAQ `
` items. I need to check if they are duplicated or overlapping with H2/H3. The FAQ items are:
* “How old is Calista Flockhart?” (separate)
* “Is Calista Flockhart still acting?” (matches H3 “Is she still acting?” – this might be an overlap, but rule says “Sub-questions that narrow a heading are KEPT.” Here, “Is Calista Flockhart still acting?” is a broader version of H3 “Is she still acting?” but it is a valid standalone FAQ. It’s not >85% lexical overlap with its H2 “What does Calista Flockhart do now?” so I will keep it.
* “What is Calista Flockhart’s net worth?” (separate, answerable)
* “Is Calista Flockhart a millionaire?” (separate)
* “Who is Calista Flockhart married to?” (separate)
* “Does Calista Flockhart have a biological child?” (separate from H2 “Does Harrison Ford have a child…”, but here it’s about Calista. Rule: “Does Calista Flockhart have a biological child?” vs H2 “Does Harrison Ford have a child with Calista Flockhart?” – these are different questions. Keep.
* “Does Calista Flockhart have a child with Harrison Ford?” (separate)
* “Does Harrison Ford have a child with Calista Flockhart?” (separate)
The FAQ list has 7-8 items, well above the minimum of 5. No removals needed.
7. **GATE 6 — LINK HYGIENE + DIVERSITY:**
* I see `href=”https://abcnews.com/…”` and other URLs. I need to check for `domain.com` and internal links. I see `https://policyline.uk` mentioned in the JSON-LD but not in the body HTML. I need to check internal links (site domain or `/`). I see no internal links in the body, but the contract said `internal_links_plan` with a count of 4. Wait, the raw input doesn’t contain those internal links. The raw input has a mix of external links. I need to check for any bare `` without `target` or `rel`. The external links have `target=”_blank” rel=”noopener nofollow”` which is good.
* I need to check for `example.com` in the JSON-LD. The contract says `mainEntityOfPage` should use `website` + `slug`. The raw HTML has `https://policyline.uk/blog/calista-flockhart-facts-biography/` which is good. I will update `example.com` if present. It is not.
* Link diversity: I see domains like `townandcountrymag.com`, `businessinsider.com`, `people.com`, `goodhousekeeping.com`, `infoseemedia.com`, `abcnews.com`. That’s 6 domains. Good.
8. **GATE 7 — JSON-LD:** I need to verify the JSON-LD blocks.
* There are two `
Wait, the raw input provided in the user message contains the article HTML *and then* a P1 contract JSON *and then* research notes. But the `GATE 7` instruction says “JSON-LD (INSIDE
, overwrite placeholders) … Keep exactly two
Wait, the raw input in the prompt ends with `
` for FAQ, then an H3, then paragraphs, then the n24-tldr, then the JSON-LD scripts. But the raw input ALSO has the P1 contract JSON and research notes after the closing `
` tag? Let me re-read the raw. The raw input in the user message has:
…
Research notes: {“topic”:”Calista Flockhart…”,”language”:”en”,…}
So the actual article ends with `
`, and the “Research notes” is a separate object in the user message. So the article HTML is clean. Wait, no. Let me look at the raw again. I see a `”question_source”: “paa”` inside the raw HTML? Let me scroll down the raw input. It goes:
