Jo Bergum - Distinguished Engineer, Yahoo! Vespa - Journey of Vespa from Sparse into Neural Search

[@TheEtrepreneur]

the WHY question was super cool. You can tell someone understands (and amplifies for the rest) a subject when they reduce a whole trend to "a feature". High level jedi stuff.

[@VectorPodcast]

Thank you for validating my selection of the WHY question -- this is the moment of self-reflection, and reflection on the industry / trend -- happy to see my guests pick it up / ignite / open up.

[@avinashronanki2382]

Hello Dmitry and Kristian, Thank you for the great talk.

[@VectorPodcast]

Hi Avinash, thank you! Were there some topics that you liked the most? :)

[@avinashronanki2382]

@@VectorPodcast 1. Reciprocal Rank Fusion 2. "Phrases search" Would love to hear more about these topics

[@VectorPodcast]

@@avinashronanki2382 Thanks for sharing! You might find the paper "Using a Reciprocal Rank Fusion method to fuse the results of the sparse and the dense retrieval results" in the show notes useful. For phrase search -- I believe you can only implement it using BM25 keyword approach reliably.