💾 Archived View for bbs.geminispace.org › s › AuraGem › 5343 captured on 2023-12-28 at 16:12:45. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2023-11-04)
-=-=-=-=-=-=-
Is there any search engine in the Gemini space that implements the Page Rank algorithm as in Brin, Page, "The Anatomy of a Large-Scale Hypertextual Web Search Engine" (1998), or something similar? The Gemini space appears not to contain manipulated content ("SEO"), so PR could work pretty well.
Sep 15 · 3 months ago · 👍 clseibold
🚀 clseibold · Sep 15 at 13:26:
Yeah, some use HITS/SALSA, which is very similar to PageRank, and I think one uses PageRank. AuraGem and geminispace.info are the only ones that don't use link analysis, afaik.
None of them use all of SEO, because the only SEO type of thing in Gemini is links (which is still a problem, due to the Tightly Knit Community effect and link spam). They do all rank on FTS (Full Text Search) in various way though so that search results are relevant (AuraGem uses FTS for metadata only, atm, but this will change so that it will do FTS on text page contents soon).
I can get specifics about which ones use which algorithms, just give me a second to look at each Search Engine.
Btw, I have an article that talks a lot about the different algorithms, how they are different or similar, and the problems with searching in general:
— auragem.letz.dev/devlog/20220722.gmi
🚀 clseibold · Sep 15 at 13:45:
Ok, I have looked up each search engine that I'm aware of within Geminispace:
Gemplex uses PageRank
— https://git.sr.ht/~elektito/gemplex
Kennedy describes its algorithm as a "PageRank-derived algorithm"
— kennedy.gemi.dev/docs/features.gmi
TLGS switched from HITS to SALSA on 2022-02-02.
And I'm not sure what ranking geminispace.info uses, but I don't think it does link analysis, afaik. One would have to look more in its source. AuraGem Search doesn't do link ranking (like PageRank or SALSA/HITS), but does try to rank based on content relevancy (using FTS and prioritizing title matches over url matches or audio artist metdata matches... that type of thing)
@clseibold, thank you for your detailed explanation.