Note that there's an easy thing here, automatically creating BibTex (which you can easily do by hand, eg the content of that MathOverflow screenshot), which seems like an easy feature for the LW team to add

And the hard thing, of actually making something that eg Google Scholar crawls, and where the various infrastructure built atop citation graphs works, which I have no clue how to do.

The first one existing feels mildly useful, but significantly less useful without the second thing.

Reply

[-]gwern3y3410

You don't need BibTex, you can get away with a small set of <meta> HTML tags. This is how I get gwern.net into Google Scholar. I'm not quite sure which of these are necessary (I copied from Michael Nielsen's websites which I knew showed up) but the ones that seemed to eventually make it work were

+    <meta name="citation_title" content="$title$">
+    <meta name="og:title" content="$title$">
+    <meta name="twitter:title" content="$title$">
+    <meta name="citation_author" content="$author$">
+    <meta name="og:site" content="gwern.net">
+    <meta name="og:type" content="article">
+    <meta name="citation_publication_date" content="$created$">
+    <meta name="citation_fulltext_html_url" content="https://www.gwern.net$url$">
+    <meta name="og:url" content="https://www.gwern.net$url$">
+    <meta name="citation_fulltext_world_readable" content="">

I didn't do anything else I recall to appease the GS gods. (I do do other things like periodically 'merge' citations together to me where GS thinks they may be to different works, and check manually each hit in GS for 'gwern' to find all the users of things like the DNM Archives and jailbreak them.)

The crawl part doesn't seem to be a problem, I think GS piggybacks on G. It then does its own scraping/parsing for 'citations', presumably.

Reply

Moderation Log