Sharing Japanese Technical Information Beyond Japan: Why and How DevelopersIO Started Automatic Translation with Bedrock

Sharing Japanese Technical Information Beyond Japan: Why and How DevelopersIO Started Automatic Translation with Bedrock

DevelopersIO has launched automatic English translation of articles using Amazon Bedrock and Claude. This post covers canonical/hreflang metadata for Google Search and ranking tests run from EC2 in North America.
2026.01.14

This page has been translated by machine translation. View original

Introduction

Starting at the end of 2025, DevelopersIO began offering an automatic English translation feature for articles using generative AI.

We want to deliver technical articles written in Japanese to developers in English-speaking regions, and to raise the international recognition of Classmethod and our authors. With this background, we are starting this initiative as a small start.

This article introduces the specifications and features of DevelopersIO's English translation feature.

Specifications of the English Translation Feature

Displaying English Versions of Articles

Articles that are subject to automatic translation display a "文/A" icon, allowing you to select a language.

CodeCommit_back_jp

When you select "English" from the translation button, an English article whose URL path contains "/en/" is displayed.

CodeCommit_back_en

  • At the top of a translated article, a notice stating that it is an automatic translation and a link to the original article (View original) are automatically inserted.

https://dev.classmethod.jp/articles/aws-codecommit-returns-to-ga/

Opt-in by Authors

The DevelopersIO English translation feature currently uses an opt-in approach by authors, and only targets articles for which automatic translation has been requested.

Auto-Translation Languages

By having authors themselves decide whether to enable translation, we aim to respect the authors' wishes while rolling out the feature gradually.

Technical Features

For translation, we use Claude series models on Amazon Bedrock.

We take care to prevent information loss caused by machine translation. In particular, we have put effort into handling code blocks and similar elements, where loss tends to occur easily in technical articles.

We hope to publish an article explaining the implementation details, written by the person in charge of the implementation, at a later date.

SEO Support: Normalizing Meta Information

So that translated articles are evaluated correctly by search engines and crawlers, we normalize and attach meta information ahead of the automatic translation.

Main Meta Information Items

Item Description
canonical_url Canonical URL (explicitly specifies the original Japanese article)
alternate_urls URLs by language (hreflang: x-default, en, etc.)
author Information about the actual author (from the E-E-A-T perspective, clearly indicates that it was written by a human)
og:title / og:description Translated title and description for social media sharing
structured_data Structured data conforming to Schema.org (Article type)

Using hreflang and canonical Together

We declare the URL of the original Japanese article with canonical, and the URL of the English translated article with the alternate tag (hreflang). This is intended to let search engines correctly understand the relationship of "which is the original and which is the translated version."

Implementation Example

This is an example of the meta information attached to an actual translated article.

{
  "basic_meta": {
    "canonical_url": "https://dev.classmethod.jp/articles/aws-codecommit-returns-to-ga/",
    "alternate_urls": {
      "x-default": "https://dev.classmethod.jp/articles/aws-codecommit-returns-to-ga/",
      "en": "https://dev.classmethod.jp/en/articles/aws-codecommit-returns-to-ga/"
    },
    "author": "suzuki.ryo"
  },
  "structured_data": {
    "@type": "Article",
    "headline": "AWS CodeCommit is fully back to GA...",
    "author": {
      "@type": "Person",
      "name": "suzuki.ryo",
      "url": "https://dev.classmethod.jp/author/suzuki-ryo/"
    },
    "about": [
      { "@type": "Thing", "name": "AWS CodeCommit" },
      { "@type": "Thing", "name": "AWS" }
    ]
  }
}

UI/UX for English Readers

For translated articles, we provide the following multilingual UI.

English Display of Menus

  • Header
    Header

  • Footer
    Footer

The side menu and event announcements in Japanese are omitted in English articles.

Measuring Effectiveness

Verifying SEO Effects

To verify the SEO effects of translated articles, we checked Google search results in the browser of an EC2 instance (English Windows environment) in North America (California region).

We confirmed that, for searches using AWS-related terms, multiple translated articles appear on the first page of Google search results.

  • "CodeCommit back": The article introducing CodeCommit's full return to GA appeared 7th on the first page of Google search results.

codecommit_back

https://dev.classmethod.jp/articles/aws-codecommit-returns-to-ga/

  • "alb target optimizer": The article introducing the November 2025 ELB update that made it possible to control the maximum number of concurrent requests appeared 5th on the first page.

alb_target_optimizer

https://dev.classmethod.jp/articles/try-aws-alb-target-optimizer/

Future Outlook

For the time being, we will only provide Japanese-to-English translated articles. We will continue to improve translation quality through evaluating and adopting the latest models and optimizing prompts.

Expansion to Classmethod's overseas offices (Korea, Thailand, Germany, etc.), opening portal pages in each country's language, and translating text within images will be considered based on the response and needs following this release.

Please send requests for translated articles or comments on translation content via social media or our contact channels. We look forward to feedback from readers that will lead to future improvements.

Summary

The DevelopersIO English translation feature has been launched as a first step toward delivering Japanese technical know-how to the English-speaking world, and widely to the world through LLMs (large language models).
As a step toward global technical information sharing, we hope this feature will allow us to deliver valuable information to more people.


AI白書2026 配布中

クラスメソッドが独自に行なったAI診断調査をもとに、企業のAI活用の現在地を調査レポートとしてまとめました。企業規模別の活用度傾向に加え、規模を超えてAI活用を進める企業に共通する取り組みまで、自社の現在地を捉えるためのヒントにぜひ。

AI白書2026

無料でダウンロードする

Share this article

AWSのお困り事はクラスメソッドへ