Romain Deffayet<p>One can&#39;t judge a click model only by how well it ranks documents, we also need to make sure it actively identified and removed biases hidden in the logged data. </p><p>That&#39;s what we showed in our recent <a href="https://sigmoid.social/tags/SIGIR23" class="mention hashtag" rel="tag">#<span>SIGIR23</span></a> paper with Philipp Hager, Jean-Michel Renders and Maarten de Rijke.</p><p><a href="https://arxiv.org/abs/2304.09560" target="_blank" rel="nofollow noopener noreferrer" translate="no"><span class="invisible">https://</span><span class="">arxiv.org/abs/2304.09560</span><span class="invisible"></span></a></p><p><a href="https://sigmoid.social/tags/PaperThread" class="mention hashtag" rel="tag">#<span>PaperThread</span></a> <a href="https://sigmoid.social/tags/ULTR" class="mention hashtag" rel="tag">#<span>ULTR</span></a> <a href="https://sigmoid.social/tags/ClickModels" class="mention hashtag" rel="tag">#<span>ClickModels</span></a> <a href="https://sigmoid.social/tags/IR" class="mention hashtag" rel="tag">#<span>IR</span></a></p>