Translation suggestions: Topic-based & Community-defined lists/Metrics

Project Overview

This project improved the article suggestion recommendation API within the Content translation tool and introduced a customizable article suggestion filters. This feature enabled editors to discover and translate vital articles on various topic areas as well as community-curated article lists for Wikiprojects and Campaigns, allowing the Language and Product Localization team to test whether this initiative enhances content coverage in Wikipedia projects. This work aligned with the WMF 2024-25 Annual Plan Objectives, specifically under the Equity Goal of providing technological support to help communities bridge knowledge gaps. For more detailed information about the project, please refer to the following pages:

Key Metrics

All metrics reported herein reflect final project outcomes as of June 2025, marking the completion of the test project period.

Content creation and coverage

The following metrics detail the translation activity associated with the article suggestion feature:

A total of 200 articles were translated using this custom suggestions feature.
Translations were initiated into 24 different target languages.
Arabic Wikipedia (arwiki) had the highest adoption rate, accounting for approximately 40% of all translations.
On average, 25 unique users accessed the feature daily. This number does not account for regular users.

Distribution of Translated Articles by Primary Topic:

Culture-miscellaneous: 67 articles (34.4%) - Largest category
Geography: 43 articles (22.1%)
Culture-biography: 33 articles (16.9%)
STEM: 29 articles (14.9%)
History and Society: 23 articles (11.8%)

Feature Utilization Breakdown

Topic areas: 42% - Most popular suggestion source
Search results (within custom suggestions menu): 30%
Collections (Community-defined lists): 23%

Analysis

Adoption Success

The project achieved significant adoption, with 200 completed translations across 29 languages. This success highlights the effectiveness of curated translation suggestions. The dominant usage of Arabic Wikipedia, making up 40% of all translations, suggests either strong community engagement or effective alignment of topics with the interests of editors.

Content Diversity

The distribution of topics indicates that the feature successfully supported diverse content creation. Cultural topics accounted for 51.3% (combined from Culture-misc and Culture-biography), making them the most popular, followed by geographical content at 22.1%. This pattern aligns with typical trends observed in knowledge gap initiatives.

Feature Preference

A preference for topic-based suggestions was observed at 42%, while only 23% preferred community-curated collections. This indicates that algorithmically categorized topics may be more discoverable or practical than curated lists. Although it is not clear why community-curated collections are less favored, several factors may have influenced their usage, such as:

The popularity of the campaign initiatives included in the collection
The regional coverage of the campaign
The naming of the campaign within the collection
Timing of the deployment

User Engagement

An average of 25 daily unique users indicates focused but consistent engagement with the feature, suggesting it attracted dedicated translators rather than casual usage.

Conclusion

The Translation suggestions: Topic-based & Community-defined lists project successfully validated the hypothesis that curated translation suggestions can increase content coverage. The feature demonstrated:

Strong multicultural adoption across 24 language communities
Balanced content creation spanning major knowledge areas
Sustainable user engagement with consistent daily usage
Preferences among different categories of suggestions

Based on these results, a key recommendation for future development is to investigate ways to enhance the discoverability and adoption of community-defined collections.

_{Project Duration: June 2024 - June 2025}

_{Reporting Period: Final metrics as of project completion}