From 83317797b187fcdaf407eb223a492a23a0919b82 Mon Sep 17 00:00:00 2001 From: Abe Tomoaki Date: Wed, 19 Aug 2026 18:22:56 +0900 Subject: [PATCH] users: Add English text Related: GitHub GH-465 --- _po/ja/users/index.po | 52 +++++++++++++++++++++++++++++++++++++++++++ users/index.md | 13 +++++++++++ 2 files changed, 65 insertions(+) diff --git a/_po/ja/users/index.po b/_po/ja/users/index.po index 103835b7..ca5667fb 100644 --- a/_po/ja/users/index.po +++ b/_po/ja/users/index.po @@ -359,6 +359,58 @@ msgstr "" "IvorySQL内でPGroongaのインデックスを作成し、シンプルなSQLクエリを通じて簡単に" "高速な全文検索を利用できるため、開発や運用のハードルを大幅に下げられます。" +msgid "## kokkai-data (国会議事録検索) {#kokkai-data}" +msgstr "## 国会議事録検索 {#kokkai-data}" + +msgid "" +"[kokkai-data (国会議事録検索)](https://kokkai-data.com) is a site for full " +"text search of the minutes of Japan's National Diet. It's developed and " +"operated by an individual. It contains about 9.6 million speeches from May " +"1947 to the present, and you can trace speakers, meetings, and dates from " +"keywords." +msgstr "" +"[国会議事録検索](https://kokkai-data.com)は、国会の会議録を全文検索できるサイ" +"トです。個人で開発・運用しています。1947年5月から現在までの発言約960万件を収" +"録していて、キーワードから発言者・会議・日付をたどることができます。" + +msgid "" +"PGroonga handles the full text search. A single index is built on the speech " +"text column using `TokenBigramSplitSymbolAlphaDigit` and " +"`NormalizerNFKC150`, and searches are done with `&@~`. Searches across about " +"9.6 million records return in tens of milliseconds." +msgstr "" +"全文検索はPGroongaが担当しています。発言本文のカラムに" +"`TokenBigramSplitSymbolAlphaDigit`と`NormalizerNFKC150`でインデックスを1本作" +"り、`&@~`で検索しています。約960万件に対する検索が数十ミリ秒で返ってきます。" + +msgid "" +"The reason for choosing PGroonga was that it doesn't require setting up a " +"search engine outside of PostgreSQL. Since it's operated by an individual, " +"it's hosted on a single VPS (virtual 6 cores / 8GB memory) shared with other " +"sites. There wasn't room to prepare another server for search, and I didn't " +"want to increase the number of things to back up and monitor. With PGroonga, " +"all you need is to build a single index, and the search conditions can be " +"written in ordinary SQL `WHERE` clauses, so there's no need to implement " +"filtering and sorting twice, once in the search engine and once in the " +"application." +msgstr "" +"PGroongaを選んだ理由は、PostgreSQLの外に検索エンジンを立てなくてよいことで" +"す。個人での運用なので、VPS1台（仮想6コア / メモリ8GB）に他のサイトと相乗りで" +"載せています。検索用のサーバーをもう1台用意する余裕はなく、バックアップや監視" +"の対象を増やしたくもありませんでした。PGroongaならインデックスを1本作るだけで" +"済み、検索条件も通常のSQLの`WHERE`に書けるので、絞り込みや並べ替えを検索エン" +"ジン側とアプリ側で二重に実装する必要がありません。" + +msgid "What I've learned from running it is summarized in articles." +msgstr "運用して分かったことは記事にまとめています。" + +msgid "" +"* [pg_total_relation_sizeが20GBって言うから信じてたら、実際は56GB使ってた]" +"(https://zenn.dev/sato_ken/articles/667459027f2025)\n" +"* [「Ｇ７」も「G7」も正規化するとg7になる。それでも検索結果は0件だった]" +"(https://zenn.dev/sato_ken/articles/651faad6e137ac)" +msgstr "" + msgid "## (Send us your service name)" msgstr "## （サービス名を教えてください）" diff --git a/users/index.md b/users/index.md index 2eda74b0..a4782a3d 100644 --- a/users/index.md +++ b/users/index.md @@ -88,6 +88,19 @@ PGroonga also scales efficiently and maintains high performance even with comple The integration of PGroonga with IvorySQL was very smooth. Its exceptional search speed and high scalability allow us to deliver an experience for our users that rivals proprietary search engines, without the need for additional complex architecture. Users simply need to create a PGroonga index within IvorySQL to enjoy fast full-text search through simple SQL queries, significantly lowering the barrier to development and operational costs. +## kokkai-data (国会議事録検索) {#kokkai-data} + +[kokkai-data (国会議事録検索)](https://kokkai-data.com) is a site for full text search of the minutes of Japan's National Diet. It's developed and operated by an individual. It contains about 9.6 million speeches from May 1947 to the present, and you can trace speakers, meetings, and dates from keywords. + +PGroonga handles the full text search. A single index is built on the speech text column using `TokenBigramSplitSymbolAlphaDigit` and `NormalizerNFKC150`, and searches are done with `&@~`. Searches across about 9.6 million records return in tens of milliseconds. + +The reason for choosing PGroonga was that it doesn't require setting up a search engine outside of PostgreSQL. Since it's operated by an individual, it's hosted on a single VPS (virtual 6 cores / 8GB memory) shared with other sites. There wasn't room to prepare another server for search, and I didn't want to increase the number of things to back up and monitor. With PGroonga, all you need is to build a single index, and the search conditions can be written in ordinary SQL `WHERE` clauses, so there's no need to implement filtering and sorting twice, once in the search engine and once in the application. + +What I've learned from running it is summarized in articles. + +* [pg_total_relation_sizeが20GBって言うから信じてたら、実際は56GB使ってた](https://zenn.dev/sato_ken/articles/667459027f2025) +* [「Ｇ７」も「G7」も正規化するとg7になる。それでも検索結果は0件だった](https://zenn.dev/sato_ken/articles/651faad6e137ac) + ## (Send us your service name) (Send us your service description, how do you use PGroonga and why did you choose PGroonga.)