From 3928df3ab312b0b0492d7417ab048b5339cbfed9 Mon Sep 17 00:00:00 2001 From: Rohan Kumar Date: Fri, 24 May 2024 08:27:17 -0400 Subject: [PATCH] Elaborate on word substitutions --- content/posts/search-engines-with-own-indexes.gmi | 2 +- content/posts/search-engines-with-own-indexes.md | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/content/posts/search-engines-with-own-indexes.gmi b/content/posts/search-engines-with-own-indexes.gmi index 5c57319..85a878e 100644 --- a/content/posts/search-engines-with-own-indexes.gmi +++ b/content/posts/search-engines-with-own-indexes.gmi @@ -469,7 +469,7 @@ Professional critics often work anonymously because personalization can damage t Some engines, like Kagi and the Ask.com family of engines, have unique-looking results from external indexes. Unique results alone don't always imply independence, as an engine could alter ranking or add filters (something that very few engines are permitted to do; Google and Microsoft generally impose a strict ToS forbidding modification). -Another possible indicator I look for is word substitutions. Returning the same results for "matza gebrent" and "matzo brei" implies a deep understanding of related food topics. Google and Bing return nearly identical results for the two queries, but engines like Mojeek return entirely different results. I often compare an engine's word substitutions to see if they're similar to another engine's, and see how many results from the top 20 are not present in the top 30-40 on other engines. I have a working list of other word substitutions I test. +Another possible indicator I look for is word substitutions. Nearly every engine supports substitutions for verb tense or singular/plural word forms, but more advanced semantic substitutions are less common. Returning the same results for "matza gebrent" and "matzo brei" requires a deep understanding of related food topics. Google and Bing return nearly identical results for the two queries, but engines like Mojeek return entirely different results. I often compare an engine's word substitutions to see if they're similar to another engine's, and see how many results from the top 20 are not present in the top 30-40 on other engines. I have a working list of other word substitutions I test. ### Caveats diff --git a/content/posts/search-engines-with-own-indexes.md b/content/posts/search-engines-with-own-indexes.md index e0bf275..9955b0e 100644 --- a/content/posts/search-engines-with-own-indexes.md +++ b/content/posts/search-engines-with-own-indexes.md @@ -494,7 +494,7 @@ Professional critics often work anonymously because personalization can damage t Some engines, like Kagi and the Ask.com family of engines, have unique-looking results from external indexes. Unique results alone don't always imply independence, as an engine could alter ranking or add filters (something that very few engines are permitted to do; Google and Microsoft generally impose a strict ToS forbidding modification). -Another possible indicator I look for is word substitutions. Returning the same results for "matza gebrent" and "matzo brei" implies a deep understanding of related food topics. Google and Bing return nearly identical results for the two queries, but engines like Mojeek return entirely different results. I often compare an engine's word substitutions to see if they're similar to another engine's, and see how many results from the top 20 are not present in the top 30-40 on other engines. I have a working list of other word substitutions I test. +Another possible indicator I look for is word substitutions. Nearly every engine supports substitutions for verb tense or singular/plural word forms, but more advanced semantic substitutions are less common. Returning the same results for "matza gebrent" and "matzo brei" requires a deep understanding of related food topics. Google and Bing return nearly identical results for the two queries, but engines like Mojeek return entirely different results. I often compare an engine's word substitutions to see if they're similar to another engine's, and see how many results from the top 20 are not present in the top 30-40 on other engines. I have a working list of other word substitutions I test. ### Caveats