We just wrapped up our webinar on 𝗢𝗻𝗲 𝗗𝗼𝗰𝘂𝗺𝗲𝗻𝘁, 𝗠𝗮𝗻𝘆 𝗩𝗲𝗰𝘁𝗼𝗿𝘀: 𝗘𝗻𝘁𝗶𝘁𝘆- ...
Milvus(@milvusio) · 人工智能
We just wrapped up our webinar on 𝗢𝗻𝗲 𝗗𝗼𝗰𝘂𝗺𝗲𝗻𝘁, 𝗠𝗮𝗻𝘆 𝗩𝗲𝗰𝘁𝗼𝗿𝘀: 𝗘𝗻𝘁𝗶𝘁𝘆- 𝗮𝗻𝗱 𝗘𝗹𝗲𝗺𝗲𝗻𝘁-𝗟𝗲𝘃𝗲𝗹 𝗦𝗲𝗮𝗿𝗰𝗵 𝘄𝗶𝘁𝗵 𝗠𝗶𝗹𝘃𝘂𝘀 𝗦𝘁𝗿𝘂𝗰𝘁𝗔𝗿𝗿𝗮𝘆. If you missed it live, catch up with the recording and slides below. 𝗥𝗼𝗵𝗶𝘁 𝗡𝗶𝗷𝗵𝗮𝘄𝗮𝗻, Senior Solutions Architect at Zilliz, showed how StructArray keeps multiple chunk embeddings and their metadata within a single document, enabling both entity-level and element-level search. You can retrieve the parent article through a matching passage or pinpoint the exact chunk and its position within the article. The walkthrough covered: • Comparing title and passage searches • Filtering individual chunks and grouping results by document • Combining matching passages with surrounding context to build answers with citations • Reusing existing embeddings and choosing when to use a flat schema ❤️ Thanks to everyone who joined and brought questions about PDF ingesti youtu.be/OT9XKJAcDGI d retrieval design to th …054828.fs1.hubspotusercontent-na1.net/hubfs/24054828… he recording: https://t.co/DJpKoL3tdD 📊 Download the slides: https://t.co/UhLYp3NqqD 💬 0 🔄 0 ❤️ 1 👀 65 ⚡ Powered by xgo.ing