Comuni Italia: quanti sono, popolazione e superficie

– Card 4: bullets have no sources either? First: “Nuove fusioni potrebbero ridurre ulteriormente il numero di comuni, specialmente al di sotto dei 5.000 abitanti” – no source. Second: “L’aggiornamento Istat 2026 fornirà dati definitivi su popolazione e superficie” – no source. For these, we need sources. Research notes have a claim about 2026: “Ultimo aggiornamento: 7.896 comuni” but source? Not directly. Possibly from Istat? The contract says “must_cover_claims” for that H2 includes “Il numero totale di comuni in Italia è 7.896 (dato aggiornato al 2026).” So that is a claim. Research notes have no source for 2026 prediction. The input later in the article says “Secondo le stime dell’Istat, il numero totale di comuni dovrebbe attestarsi a 7.896 entro il 2026” and that has a link to Istat PDF. So for the bullet in card 4 about “L’aggiornamento Istat 2026”, we can add the same source. But we need to check if it’s in card 4. Yes, second bullet of card 4. We’ll add Istat referring to the same PDF. For the first bullet about new fusioni, we have no source, but it’s speculative. It can stay without source? Gate says every li must have named source anchor. If no source, we must remove. But it’s a speculation. Perhaps we can move it to a “rumor-list” but there is no separate rumor section. We’ll need to either find a source or remove. Since it’s reasonable, we could keep it as is and note it’s uncertain? But strict. I’ll check research notes: there is a claim “Le fusioni future di comuni non sono confermate” in the uncertain list. So we could change the bullet to reflect uncertainty but still need a source? The uncertain list itself is from the article? Actually the article has a later section “Cosa resta incerto” that mentions that. So we could remove that bullet from card 4 and reference that section. But to keep the card, we can rephrase to include a source? Not possible. Alternatively, we can leave it as is and apply Gate 11 later? Let’s proceed and see.

– The timeline: The section “Segnale temporale” has three bullets with sources? 1861 no source, 2008 has source Citallia, 2023 has Pagella. We’ll remove the 1861 bullet as no source. Or we can add a source from Wikipedia maybe? The research notes mention “Wikipedia – Comuni d’Italia per superficie” but not for 1861. I think it’s safer to remove it. But the card is part of snapshot block, and we already removed one from card 3. We’ll remove the 1861 bullet from card 3? Actually card 3 is “Segnale temporale” with three bullets. If we remove the 1861, we have two left. That’s okay.

– The key insights section (confirmed facts and incognite) – the lists have sources? Confirmed list has sources on each item. Incognite list has no sources on the first bullet, second bullet has no source, third has no source. But these are incognite – uncertainty. They may not require a source? The gate says “every

– Also in the snapshot block, card 2 has a bullet with source, card 4 we’ll add source for the second bullet.

Gate 4: Fact lock vs verified_facts. The input has no verified_facts provided (empty JSON). So no changes.

Gate 5: FAQ dedup. Need to ensure FAQ items do not overlap >85% with H2/H3. The FAQ items are: “Qual è il comune italiano con la popolazione più alta?” (H2: “Quali sono le 10 città più grandi d’Italia?” – different), “Qual è il comune italiano con la superficie più estesa?” (H2: “Quali sono le città più grandi d’Italia per superficie?” – close but not identical), “Quanti comuni ci sono in Lombardia?” (H2: “Quanti comuni ci sono in Italia?” – not the same), “Quali sono i comuni italiani con meno di 1.000 abitanti?” (not in H2), “Come si calcola la popolazione di un comune italiano?” (not in H2), “Qual è la differenza tra comune e provincia?” (H2: “Quante province ci sono in Italia?” – different), “Quanti comuni italiani hanno oltre 100.000 abitanti?” (H2: “Quali sono le 10 città più grandi d’Italia?” – partially overlapping but not identical). So no dedup needed. Keep all 7.

Gate 6: Link hygiene
– Internal links: two provided in input: and . They are internal (same domain). Remove target and rel? The gate says for internal: strip target/rel. So remove target=”_blank” and rel=”noopener nofollow” from these.
– External links: keep target=”_blank” rel=”noopener nofollow”. Already present. Also ensure each external URL at most once. The article has multiple links to the same PDF: https://www.istat.it/storage/ASI/2024/capitoli/C01.pdf appears multiple times. That’s allowed? The gate says “Each external URL at most once.” That means we cannot have duplicate links to the same URL. We need to deduplicate. We’ll replace subsequent occurrences with plain text or remove the link. But we need to keep citations. We can keep the first occurrence with link, and for others we can use the same anchor text but without href? That would violate. Better to use a different source? Or we can consolidate references. Since it’s the same PDF, we can leave only the first link and for subsequent references, remove the hyperlink but keep the text with a note “ibid.” or just the source name without link. But the gate says max 3 per domain, not per URL. So duplicate URL is fine as long as not repeated more than 3 per domain? Actually “Each external URL at most once.” So we must ensure each unique URL appears only once. The PDF URL appears many times. We’ll replace all duplicates with plain text (source name) but keep the first occurrence. Let’s count: intro, snapshot card 1, card 2, table rows, h2 sections, etc. Many. We’ll need to remove all but the first occurrence. We’ll do that. Also for other URLs like Wikipedia, Tuttitalia, etc., they appear multiple times. Deduplicate similarly: keep first occurrence as link, subsequent ones as plain text with source name. But careful: we must not remove all links; we need at least 10 external links overall from different domains. The article has plenty. After dedup, we’ll have one link per URL. That’s fine.