{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/b4c322552955434a9059c17e22438e30\" frameborder=\"0\" width=\"1720\" height=\"1290\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1290,"width":1720,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1290,"thumbnail_width":1720,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/b4c322552955434a9059c17e22438e30-f59c0f23395629df.gif","duration":179.2307,"title":"Improving Document Relevance in Retrieval Systems","description":"In this video, I'm discussing the need to determine precision and recall for our retrieval system, focusing on whether documents are relevant or irrelevant. I emphasize the importance of having annotated documents to create a golden dataset, which will help us evaluate the effectiveness of our search system. Currently, I can only assess the entire search result, but I want to be able to annotate documents individually and manage metadata for relevance. I also highlight the need for a way to add or remove documents based on false negatives. If anyone has insights on how to streamline this process, I would greatly appreciate your input."}