LLMs -> Semantic HTML Gets Hot Again?
“Use Semantic HTML and Clear Tagging: Ensure that headings are marked with <h1>, <h2>, ... tags in logical hierarchy, and use <p> for paragraphs, <code> or <pre> for code examples, and lists for step-by-step guides. Meaningful HTML structure helps crawlers (and by extension LLMs that rely on web data) extract the content hierarchy."
So the far too common #HTML <div> soup considered #AI harmful. Great. Better markup helps #a11y but also… 😬
ThomasAPowell
npub10e…jm7y8
2025-03-31 18:11:14 UTC
Author Public Key
npub10ewr692fqqp2u36w2pkzj82y7gm9uj27nrudlunx8e54u4xtj54shjm7y8Seen on
wss://relay.ditto.pubPublished at
2025-03-31 18:11:14 UTCEvent JSON
{
"id": "a7ff876d2e2d94028abeb1f07213b8aff66e2c1c22325b0da3b519625d3331e9",
"pubkey": "7e5c3d15490002ae474e506c291d44f2365e495e98f8dff2663e695e54cb952b",
"created_at": 1743444674,
"kind": 1,
"tags": [
[
"t",
"html"
],
[
"t",
"ai"
],
[
"t",
"a11y"
],
[
"proxy",
"https://fosstodon.org/users/thomasapowell/statuses/114258390172368443",
"activitypub"
]
],
"content": "LLMs -\u003e Semantic HTML Gets Hot Again?\n\n“Use Semantic HTML and Clear Tagging: Ensure that headings are marked with \u003ch1\u003e, \u003ch2\u003e, ... tags in logical hierarchy, and use \u003cp\u003e for paragraphs, \u003ccode\u003e or \u003cpre\u003e for code examples, and lists for step-by-step guides. Meaningful HTML structure helps crawlers (and by extension LLMs that rely on web data) extract the content hierarchy.\"\n\nSo the far too common #HTML \u003cdiv\u003e soup considered #AI harmful. Great. Better markup helps #a11y but also… 😬",
"sig": "e9510af07c13c2ebb1f7e5bbb8ed0ecdaf5c597a165d7d5755933120128b06292bbc771076097db8c33052c0d53df76182bbaad8664db22b5a8790c9689c0a34"
}