J’ai l’impression d’avoir vu un autre message ou sujet à ce sujet ici récemment, mais je ne le trouve pas.
Lorsque je recherche « feature » ici, les résultats incluent un lien vers la catégorie Contribute > Feature.
Mais lorsque je recherche « pr-welcome », la balise pr-welcome n’apparaît pas dans les résultats.
Ce serait bien que ce soit le cas, car cela rendrait les balises plus faciles à découvrir et constituerait un moyen simple de naviguer vers une balise en particulier.
I think it does if you use pr-welcome instead of pr-welcome
Good point. Guess that solves the “quick way to navigate” use case. Would still like it to work without the # to increase discoverability.
Not sure it would be a good idea. Search is usually a tradeoff between relevancy, speed, index time/volume and query complexity ( aka special keywords/filters/options). Making tags part of the directly searchable keywords would hurt relevancy in some cases ( how do you rank a post containing pr-welcome in body versus a post tagged pr-welcome?) and probably hurt performance.
It’s easy to see a query and say this should be first and this second… but once you take all the cases in consideration it’s much harder.
How so? I’d say that a hit in a tag is almost always worth more than a hit in the post content.
Besides, if your argument applied to tags it would also apply to categories (which are apparently included in the search).
Really? What are a few dozen (or even a few hundred) tags in the sea of post content?
Je ne suggère aucun changement concernant la façon dont le classement ou la pertinence est déterminée pour les sujets et les messages. Cela pourrait mériter une enquête plus approfondie, mais ce n’est pas ce que je cherchais à suggérer.
Tout ce que je veux voir, c’est le tag pr-welcome lui-même affiché sous « afficher plus » de la même manière que le lien vers la catégorie Contribute > Feature est affiché dans la première capture d’écran.
J’aurais dû mettre de grosses flèches rouges sur la capture d’écran pour le souligner plus spécifiquement. Je suis sur mobile maintenant, donc je ne peux pas facilement les mettre à jour pour le moment, mais j’espère que cela aide à clarifier de quoi je parle.
As I said it’s a tradeoff, it’s not because it has been done on one thing that we should do it for everything (opposite being true too).
If we follow your idea “tag > content”… content match versus tag match will always be low in results, so searching a term wich also happens to be a tag will be almost impossible.
That’s exactly the edge case because of which I said
But even that could be handled by making it so that hits in tags are treated the same as hits in content unless there is at least one search term that does not match a tag. Or you could simply treat tags as content, full stop.
As I said, tradeoffs. This kind of “if” is not free. For every query we now have to match against the full list of tags each word, where before we would just have to search for a #, totally different story.
And making it part of the searchable content, is just making it slower. The sooner you can reduce the list of potential results the faster it will be, so filtering on a tag with #tag and then adding a query will be probably much faster than full search on “tag query”.
Much of this conversation seems off-topic to me.
I updated the original post to better clarify what is being asked for here.
Adding tags to the “category hit” section is totally pr-welcome agree we should add it. We will need to be careful at ordering the tags list and culling it if it is too long. (Imagine 1000 tags called spam-1 spam-2 spam-3, though lovely, it could kill off that whole UI. So show popular spam tag first)
I agree with this, can you add it to your list @neil? Just show top X matching tags, ordered by tag frequency.
