#dev 2024-11-18

2024-11-18 UTC
GuestZero_, geoffo, bterry, barnaby, lazcorp and AramZS joined the channel
#
[Joe_Crawford]
HTML validation question. https://validator.w3.org/nu/?doc=https%3A%2F%2Fartlung.com%2Farchive%2Fportfolio%2F2024%2F says "*Error*: Attribute `popovertarget` not allowed on element `button` at this point." -- but it seems as though that attribute is allowed on button. Not sure I understand the "at this point" part of that.
#
[Joe_Crawford]
what's weird is the output of the validator goes through all the attributes for button and what they are and includes popovertarget. it's weird.
[mattl] joined the channel
#
[tantek]
copying what I said in #indieweb-random regarding personal site search algorithms: I would always show results with an exact phrase before results without the exact phrase (but with all the words in the query). And then yes, more occurrences of the exact phrase should be shown earlier. And then among same number of occurrences I think I would rank by how early the phrase (first) occurs on a page. cc: capjamesg[d]
[aciccarello] and sebbu2 joined the channel
#
capjamesg[d]
[tantek] The first and second sentences describe how things are now.
#
capjamesg[d]
I like the idea of weighing posts that mention a name higher up, but I haven't seen a case where I need that yet.
#
capjamesg[d]
I like to run lots of queries and see where things aren't right, then think about what would help improve the results.
#
capjamesg[d]
Which reminds me! CMoS is #3 for tantek.com: https://jamesg.blog/search-pages/?q=tantek.com&page=1
#
capjamesg[d]
๐Ÿ˜„
#
capjamesg[d]
[edit] Which reminds me! CMoS is #3 for tantek.com: https://jamesg.blog/search-pages/?q=tantek.com&page=1
#
[tantek]
hah, hilarious
#
sknebel
[Joe_Crawford]: it just doesnt support that yet: https://github.com/validator/validator/issues/1534
#
Loqi
[preview] [mtrootyy] #1534 add popover attribute
#
[Joe_Crawford]
[tantek]++ thanks. that helps! It is incongruous that the documentation for button it spits out acknowledges the attribute and yet it doesn't validate. But good to know. ๐Ÿ™‚
#
Loqi
[tantek] has 23 karma in this channel over the last year (133 in all channels)
shoesNsocks joined the channel
#
[tantek]
sknebel++ <-- that was him not me [Joe_Crawford] ๐Ÿ™‚
#
Loqi
sknebel has 9 karma in this channel over the last year (23 in all channels)
#
[Joe_Crawford]
DERP. Much sorry. Caffeine deficiency. Thanks [sknebel]++
#
Loqi
[sknebel] has 10 karma in this channel over the last year (24 in all channels)
#
[Joe_Crawford]
[sknebel]++
gRegor, [aciccarello] and [Joe_Crawford] joined the channel
#
capjamesg[d]
I am working on information extraction in my search engine...
#
capjamesg[d]
[tantek] I wrote a blog post about how the impact of word proximity as a ranking factor https://jamesg.blog/2024/11/18/search-word-proximity/
#
capjamesg[d]
[edit] [tantek] I wrote a blog post about how the impact of word proximity as a ranking factor https://jamesg.blog/2024/11/18/search-word-proximity/
#
capjamesg[d]
TL;DR: It makes a _huge_ positive difference for some queries.
#
capjamesg[d]
(Scroll down to the "For the query โ€œall too wellโ€, BM25 without a proximity boost returns:" part)
#
Loqi
[preview] Improving search relevance with word proximity
#
Loqi
[preview] Improving search relevance with word proximity
#
[mattl]
working on a sitemap for http://Libre.fm -- even just a very basic one is going to be a huge file.... but not sure if splitting it up into multiple files serves the same purpose
[morganm] joined the channel
#
[mattl]
I hadnโ€™t considered making an XML version. I was thinking more of a human sitemap but maybe I can make both with one script.
#
[aciccarello]
I only included the top level paths in my human sitemap
troojg and ttybitnik joined the channel