<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Cold Start on Chen Kai Blog</title><link>https://www.chenk.top/en/tags/cold-start/</link><description>Recent content in Cold Start on Chen Kai Blog</description><generator>Hugo</generator><language>en</language><lastBuildDate>Fri, 09 Jan 2026 09:00:00 +0000</lastBuildDate><atom:link href="https://www.chenk.top/en/tags/cold-start/index.xml" rel="self" type="application/rss+xml"/><item><title>Recommendation Systems (14): Cross-Domain Recommendation and Cold-Start Solutions</title><link>https://www.chenk.top/en/recommendation-systems/14-cross-domain-cold-start/</link><pubDate>Fri, 09 Jan 2026 09:00:00 +0000</pubDate><guid>https://www.chenk.top/en/recommendation-systems/14-cross-domain-cold-start/</guid><description>&lt;blockquote>
&lt;p>When Netflix launches in a new country, it inherits millions of users with no history and a catalog with no local ratings. Amazon faces the same issue each time it opens a new product category. Pure collaborative filtering, the workhorse of warm-state recommendations, has nothing to compute. Techniques that make recommendations work in this scenario include: bootstrap heuristics for the first request, meta-learning after a few interactions, cross-domain transfer when a related domain is rich, and bandits to keep exploring once the model is confident. This post walks through these techniques, anchored to the papers they come from.&lt;/p></description></item></channel></rss>