` > `

Senaste artiklar

` intro
– There’s already a stats line, snapshot block, FAQ section with details, JSON-LD script
– But I notice some issues:

**GATE 14.2** — Table lead-ins: The first table `

` has a lead-in (”Här är en snabb bild över de viktigaste fakta som presenteras i den här guiden:”) — this is OK.

**GATE 14.3** — Section closers: Need to verify each H2 section ends with a takeaway. Looking at the sections:

– Section after ”Kan man se hjärtljud i vecka 6?” ends with a blockquote (`

` is closing an FAQ, but wait, let me check the structure more carefully)

Actually, looking at the raw HTML, I notice a problem – there’s unclosed `

` and `

` tags. Let me check:

Hur stor är risken för missfall i vecka 6?

Risken för missfall i vecka 6 är fortfarande relativt hög, men den minskar avsevärt efter att ett hjärtslag har bekräftats. Enligt True Ultrasound faller risken till mellan 5 och 10 procent efter ett bekräftat hjärtslag i vecka 7.

This appears to be cut off mid-way and there’s a missing closing for the first FAQ item. Let me look at the full input more carefully…

I see the input has `` which appears after some JSON, then continues with more JSON-LD. But there’s a problem: the FAQ section has a partial `

` block that’s incomplete, and the overall structure seems malformed.

Also I need to check:

**GATE 8** — The forbidden phrases list. Let me scan the body text for them:

– ”navigating the complexities” – not found
– ”stands as one of the” – not found
– ”it’s important to note” – not found
– etc.

Let me scan more carefully…

Actually, I see at the top there are research notes embedded in the JSON, and then the actual HTML starts. But there’s a problem: the ”, there’s more research notes JSON… that seems wrong.

Let me re-read the raw input more carefully:

När försvinner hjärtslagen om man får missfall?

Om ett missfall inträffar efter att hjärtslag har setts, så avstannar hjärtat vanligtvis inom några dagar. Enligt Terveyskylä / Kvinnohuset kan man ibland se tecken på att graviditeten har avstannat vid ett nytt ultraljud.

OK so the FAQ details close properly, then there’s a `` closing the JSON-LD script. Then we have research notes JSON.

Now I understand the structure. The `

` contains:
– intro paragraph
– stats line
– snapshot block
– table
– H2s with content
– FAQ section (details elements)
– JSON-LD script(s)

Now let me check the issues:

**Gate 8 — Forbidden phrases**: Scanning the Swedish text:
– ”i den här guiden” → Swedish for ”in this guide” → should be removed
– ”sammanfattningsvis” → Swedish for ”in summary” → should be removed

Wait, but the forbidding list has English phrases. Let me re-read: ”GATE 8 — TONE HYGIENE” with phrases like ”stands as one of the”, ”increasingly shape”, etc. These are all in English. The article is in Swedish.

Hmm, but the instruction also says ”STRIP any occurrence in body text (outside



Then after that:

Kan man se hjärtljud i vecka 6?

Hmm, this is getting confusing. The input seems to have the article content, then research notes outside the article.

Let me re-read the task description. The raw article HTML is provided, and I need to repair it. The research notes are in a separate structure (the JSON with claims, facts, etc.).

OK, I think the input I'm given IS the article HTML. Let me focus on what the repaired output should be.

Given the complexity of this task, let me focus on the key issues:

1. **GATE 14.2** — Table lead-in check: The table has "Här är en snabb bild över de viktigaste fakta som presenteras i den här guiden:" — this is a lead-in, but it uses "den här guiden" (= "in this guide") which might be an issue. It's not "in this guide" verbatim in English, but in Swedish it's natural. I'll keep it.

2. **GATE 14.3** — Section closers: I need to check each H2 section.

Looking at the article content (the visible text), I see these H2s:
- "Kan man se hjärtljud i vecka 6?"
- "Kan tidigt ultraljud orsaka missfall?"
- "Hur skiljer sig rekommendationerna mellan olika svenska städer?"
- "Sammanfattning"
- "Vanliga frågor"

Let me check the section "Kan man se hjärtljud i vecka 6?" — it ends with a paragraph about Flo Health and the risk of miscarriage, but actually no, let me trace through:

Looking at the raw HTML, after the "Kan man se hjärtljud i vecka 6?" H2, we have:
- Paragraph about "För att sätta saker i perspektiv..." — this continues with "risken för missfall är som högst under vecka 5–6..."

Then we have the H2 "Hur skiljer sig rekommendationerna mellan olika svenska städer?" — so the first H2 section ends there.

The last element in that section is a paragraph. The section doesn't end with a table, ul, ol, or callout — it ends with a paragraph. But is it an analytical takeaway? The last paragraph in that section is:

"Enligt en sammanställning från True Ultrasound är risken för missfall efter ett bekräftat hjärtslag i vecka 7 så låg som 5–10 procent."

This is more of a factual statement than an analytical takeaway. I might need to add a sentence.

Actually, wait. Let me look at the GATE 14.3 requirement: "Every H2 content section ends with analytical takeaway (not with a table, list, or callout). If the last element of an H2 section is a

,

    ,

      , or callout div, append a closing

      with interpretive sentence"

      So if the last element is a paragraph (not a table/list/callout), then it's fine. But if it's a callout/table/list, I need to add a paragraph.

      Looking at the section ending with "...så låg som 5–10 procent." — that's a paragraph, so it satisfies GATE 14.3. But wait, I need to check if "Det finns en diskrepans i hur tidigt ultraljud erbjuds..." paragraph is before or after the H2 "Hur skiljer sig rekommendationerna..."

      Actually, looking at the structure:
      - The "Hur skiljer sig rekommendationerna mellan olika svenska städer?" H2 is followed by a paragraph about cities and clinics.

      Hmm, I'm getting confused by the interleaved content. Let me focus on what I need to verify.

      OK, I realize I need to just do the repairs systematically. Let me go through each gate:

      **GATE 1** — Structure:
      - First child of `

      ` must be intro paragraph with class="n24-p wp-block-paragraph" itemprop="description" — ✓ It's there.
      - No leading whitespace from `

      ` — ✓ starts with `

      ` directly.

      **GATE 8** — Forbidden phrases: Need to scan Swedish text for any English AI-tells. I don't see any English AI-tells in the Swedish text.

      **GATE 9** — Quote speaker variety: The article doesn't have blockquotes with speakers - it just has paragraphs mentioning sources. So this gate may not apply. But let me check: "If all blockquotes are from same speaker" — the article has no `

      ` elements. They're all `

      ` for FAQ or regular `

      `. So GATE 9 doesn't apply.

      **GATE 10** — Research confidence: The research_confidence field shows "low". But the verified_facts are empty, and the claims in the research notes have various confidence levels. The article as written has specific claims with sources. I don't see a clear "rumor list" vs "confirmed list" to compare. I'll check the research_confidence field.

      **GATE 11** — Facts summary tiers: The verified_facts are empty `[]`, so nothing to check.

      **GATE 12** — UX:
      - Stats line: present ✓
      - Key facts table: present ✓
      - At least 2 callouts (n24-tip/n24-note/n24-warning): I need to check... I see `n24-tldr` which is a TLDR but not one of the required callout types. I need to add at least 2 callouts of type tip/note/warning.

      Looking at the article, I see `

      ` — yes! Looking at the raw HTML: `

      ` appears. And also `

      `. Let me check...

      In the raw HTML I see: `

      ` — this is the warning callout. And I see `

      ` which is a TLDR. But where's the second callout?

      Let me look for `n24-tip`, `n24-warning`, or `n24-note` divs in the content...

      I see in the FAQ section there's no callout. But looking at the main content, I need to check if there are at least 2 callouts. If not, I need to add them from the verified facts.

      Actually, looking at the input, I see two callout divs: `

      ` (a TLDR) and the warning callout in the FAQ... no wait.

      Let me look more carefully. The article has:
      1. `

      ` — not in the allowed callout list (n24-tip, n24-note, n24-warning)
      2. Is there a `n24-warning`? I see `

      ` — yes, I can see that in the raw HTML.

      But that seems to be the only one. I need at least 2.

      Looking at the raw HTML, I see:

      Bara fakta: ...

      But that's the only one visible. I might need to create another callout (e.g., a `n24-tip`) from existing claims.

      **GATE 12** also requires `FAQ with ≥5

      `. Looking at the FAQ, I count 4 items in the "Vanliga frågor" section... actually, let me count:
      1. "Kan man se hjärtljud i vecka 6?"
      2. "Hur stor är risken för missfall i vecka 6?"
      3. "Vilken vecka kan man göra ett tidigt ultraljud?"
      4. "Kan tidigt ultraljud orsaka missfall?"
      5. "Vad är skillnaden mellan tidigt ultraljud och KUB?"
      6. "När försvinner hjärtslagen om man får missfall?"

      That's 6 FAQ items, which is ≥5. ✓

      But wait, looking at the input more carefully, the first FAQ item seems malformed - the `

      ` might not have `

      `. Let me check... Looking at:

      Kan man se hjärtljud i vecka 6?

      That seems fine.

      **GATE 5 — FAQ dedup**: Check if any FAQ items duplicate H2s or H3s. Looking at the FAQ items and the H2s:
      - "Kan man se hjärtljud i vecka 6?" — duplicates the H2 "Kan man se hjärtljud i vecka 6?"
      - "Hur stor är risken för missfall i vecka 6?" — this H2 exists in the content plan but not in the article HTML (it's only in the research notes)
      - Etc.

      Wait, looking at the actual FAQ section in the HTML, all FAQ items seem unique. But I need to check if any duplicate H2s.

      Actually, looking at the FAQ questions, they seem to address broader questions not covered by H2s. The dedup rule says: "Remove FAQ ONLY if >85% lexical overlap with H2/H3 OR exact same question."

      "Hur stor är risken för missfall i vecka 6?" — this is a question, not an H2/H3 heading. But wait, in the research notes, there's an H2 "Hur stor är risken för missfall i vecka 6?" — but that's in the research notes, not in the article itself.

      Looking at the actual article H2s:
      - "Kan man se hjärtljud i vecka 6?"
      - "Hur skiljer sig rekommendationerna mellan olika svenska städer?"
      - "Sammanfattning"
      - "Vanliga frågor"

      So the FAQ item "Kan man se hjärtljud i vecka 6?" duplicates H2 "Kan man se hjärtljud i vecka 6?" — removing this FAQ item would drop it to 5 items, which is still ≥5.

      But I should keep it if possible. The rule says "Remove FAQ ONLY if >85% lexical overlap" — is it 100% match? Yes, it's the same text. But the rule says to only remove if it duplicates H2/H3. Let me re-read: "Remove FAQ ONLY if >85% lexical overlap with H2/H3 OR exact same question."

      Since "Kan man se hjärtljud i vecka 6?" is exactly the same as the H2, I should remove that FAQ item. But that would reduce the count to 5, which is still ≥5. But actually, looking at the FAQ items, there are 6 total. If I remove one, I get 5 — still OK.

      But wait, I also need to check if any FAQ items duplicate each other. Let me check:
      1. "Kan man se hjärtljud i vecka 6?" — H2 overlap
      2. "Hur stor är risken för missfall i vecka 6?" — unique
      3. "Vilken vecka kan man göra ett tidigt ultraljud?" — unique
      4. "Kan tidigt ultraljud orsaka missfall?" — unique
      5. "Vad är skillnaden mellan tidigt ultraljud och KUB?" — unique
      6. "När försvinner hjärtslagen om man får missfall?" — unique

      Item 1 duplicates H2. I should remove it. But that leaves 5, which is the minimum. Actually, the minimum is 5, so removing one to get 5 is OK.

      But wait, actually the rule says "Remove FAQ ONLY if..." — so I should remove it, but only if ≥85% overlap. It's 100% overlap (exact match). So yes, remove it. That gives us 5 FAQ items, which is the minimum required.

      Hmm, but looking at this more practically: removing it would reduce to 5 items, which is the minimum. But we could also keep it and just make sure it doesn't have exact same question. But the rule says to remove if exact same question. Maybe I should keep it but rephrase slightly? No, the rule is clear: remove if >85% lexical overlap or exact match.

      Actually, looking at the FAQ items, none of them exactly match the H2s. The H2s in the article are:
      - "Kan man se hjärtljud i vecka 6?"
      - "Kan tidigt ultraljud orsaka missfall?" (wait, is there such an H2 in the article? Let me check)

      Looking at the raw HTML (the article body), the H2s I can identify are:
      1. "Kan man se hjärtljud i vecka 6?" (from the h2)
      2. "Kan tidigt ultraljud orsaka missfall?" (this might be an H3 under a different H2, or maybe it's missing?)

      Actually, I see in the raw input the H2s listed but they're

      Relaterade artiklar

      Fler artiklar
      Robin Lindberg
      Robin LindbergReporter

      Robin Lindberg är reporter på Stadsposten och bevakar dagliga nyheter och händelser.