I have just posted a new devlog. More works needs to be done if we want to be more accurate with our conclusions, but this is a start. Hopefully it is useful.
Gemini Search Results Study, Part 1:
gemini://auragem.space/devlog/20220807.gmi
2 years ago 路 馃憤 mfoo2, freezr, moddedbear
gemini://auragem.space/devlog/20220807.gmi
@freezr I have made some clarifications in my post. If you could reread it and provide feedback, that would be useful. :) 路 2 years ago
@freezr Also, yes, most everything is an "implicit assumption" in search engines, that's just the way it is, lol. That's why we create a convention or use an existing convention of search queries so that we can deduce important information. For example, if I search "AuraGem", then I want to get back AuraGem's capsule - this is a natural language convention where you use the name of something to refer to that thing. How auragem implements this is if a term in the query has a match with a domain name, then the result is boosted because it's as if that term was a Proper Noun for the domain. Obviously all natural language has ambiguities that must be handled. 路 2 years ago
@freezr Also, one more thing. The 6 general attributes are not assumed to be desired all of the time. A user doesn't always want an *exact* url to show up, and that's fine, because what you want is dependent on what *type of search* you are doing. 路 2 years ago
@freezr So yes, "exactness" is not wanted necessarily when searching general categories of sites. Instead you want relevancy and maybe some other things. The whole point is what the search engine should do is to give you back exact urls that you want if you want exactness, like for queries like "Station", but not give you exactness when you *don't* want that, for other types of searches, so the search engine must prioritize something else.
The article only talks about what is expected by the user and categorizes types of searches based on *expectation*. How to *achieve* these results is not the goal of this article, but an article in the future. 路 2 years ago
@freezr I suppose I didn't explain it clearly enough within the article, idk, but "Exactness" is the measure of accuracy for a very specific URL that a person wanted within search results. The point is the search engine had to find a way to deduce or assume what the exact result is that the user wanted. It's accuracy is important only for *some* types of searches where exactness is wanted (like the very first type of search in my post - searching by capsule name). 路 2 years ago
Here my two cents... HTML uses <meta> tags to define contents and other stuff, on our side the specs say that we should use robot.txt[1] to help crawlers.
I haven't used any robot.txt so far and I do not know if all those search engines rely in any measure on them.
But if this would help having a better indexing of the Geminispace, this should be yelled clearly and spread all around the Geminispace, until it becomes a standard practice to have a robot.txt with a "standardized" pattern that you (devs) may suggest.
Without a robot.txt I don't know how any search engines can categorize capsules or create a cognitive map of the Geminispace... 馃 路 2 years ago
That was an interesting reading, however I have some perplexities in some criteria.
Exactness toward users: this looks to me an implicit assumption, and it may fall users searching into a category.
Continuing speaking about "categories": how are those determined?
馃 路 2 years ago