<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Toolforge on A blog</title><link>https://blog.vfiles.no/tags/toolforge/</link><description>Recent content in Toolforge on A blog</description><generator>Hugo</generator><language>en-us</language><copyright>&lt;a href="https://creativecommons.org/licenses/by-sa/4.0/deed.en" target="_blank" rel="noopener"&gt;CC BY-SA 4.0&lt;/a&gt;, except where indicated otherwise</copyright><lastBuildDate>Fri, 08 May 2026 22:50:31 +0200</lastBuildDate><atom:link href="https://blog.vfiles.no/tags/toolforge/index.xml" rel="self" type="application/rss+xml"/><item><title>Building WikiVisage: Active Learning for Wikimedia Commons</title><link>https://blog.vfiles.no/posts/building-wikivisage-active-learning-for-wikimedia-commons/</link><pubDate>Fri, 08 May 2026 22:50:31 +0200</pubDate><guid>https://blog.vfiles.no/posts/building-wikivisage-active-learning-for-wikimedia-commons/</guid><description>&lt;p&gt;&lt;a href="https://commons.wikimedia.org/wiki/Main_Page"&gt;Wikimedia Commons&lt;/a&gt; has over 100 million freely licensed media files. A lot of these are photographs of people, politicians, artists, scientists, and athletes, but most of them lack structured metadata saying &lt;em&gt;who&lt;/em&gt; is actually in the picture. That metadata is called &lt;a href="https://commons.wikimedia.org/wiki/Commons:Structured_data"&gt;Structured Data on Commons&lt;/a&gt; (SDC), and the specific property is &lt;a href="https://www.wikidata.org/wiki/Property:P180"&gt;P180 (depicts)&lt;/a&gt;.&lt;/p&gt;
&lt;p&gt;Without it, finding &amp;ldquo;all photos of Douglas Adams&amp;rdquo; means someone has to manually tag every single one. There are millions of images that need this. I wanted to see if I could make that process faster.&lt;/p&gt;
&lt;p&gt;I built WikiVisage to make that process faster. It&amp;rsquo;s an active learning tool that uses face recognition to help users classify faces in Commons images, builds a lightweight classifier from their input, and then writes the results back as structured data. It&amp;rsquo;s open source, hosted on Wikimedia Toolforge, and available at &lt;a href="https://wikivisage.toolforge.org/"&gt;wikivisage.toolforge.org&lt;/a&gt;. Anyone with a Wikimedia account can use it.&lt;/p&gt;</description></item></channel></rss>