If your recommendations are still similar in 2024, you’re doing it wrong (respectfully). Here’s a summary of what’s in and out in the world of vector databases in 2024. 1. One Type of Recommendation Is Out; a Balanced Set Is In ❌ Out: offering only 1 type of search like Top-K. Don’t get me wrong, top-K is core to vector search, but sometimes, it gives recommendations of items that are too similar, affecting recommendation quality. Just because someone listens to Adele on sad days, that doesn’t mean they want to listen to her all the time. ✅ In: The inclusion of ensures a more ✨balanced✨ set of results by allowing you to define a distance range for vector similarity. Balanced recommendations help prevent recommending things that are too similar or too disparate. range search 2. Extra Steps to Normalize Vectors Are Out, Using Cosine Similarity Is In ❌ Out: Taking an extra step 👎to normalize a vector to measure (identifying sentences or phrases that convey similar meanings to each other) or relatedness in various domains. similarity ✅ In: allowing you to easily normalize a vector in one step 👍 Cosine similarity 3. Updating Data in Multiple Steps Is Out; Using Upsert to Do It Seamlessly Is In ❌ Out: Inefficiently updating data in a database in a frustrating 😡two-step process: delete, then insert. This can’t ensure data atomicity and operational convenience. ✅ In: simplifies the update process: if data does not exist in the system, it inserts it; if it exists, it updates it. 😀 Upsert 4. The Cloud Provider Triad (AWS, GCP, and Azure) Is In ❌ Out: Vector databases not being available on the cloud provider triad: AWS, GCP, and Azure. ✅ In: Vector databases that are available on 3 major cloud providers 🌟(including GCP Marketplace) and 8 regions across North America, Europe, and Asia 👀 AKA . Zilliz Cloud A world with a vector database being available on all 3 cloud platforms is no longer imaginative. It’s true with Zilliz Cloud. If you’re interested in trying range search, upsert, and cosine similarity (on any of the 3 cloud platforms), get started . here There’s also a live webinar covering these features and more on February 1st. Register . here There will also be a live Q&A on Discord. Join us . here

This post provides insights into new product. 

Is vector database a better option for you?

Too Long; Didn't Read

Make resilience your competitive advantage

4 Ins and Outs of 2024: Vector Database Edition

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

Untitled Story

What Are Convolution Neural Networks? [ELI5]

The Noonification: Have U Been Pwned? (1/12/2023)

Goldman Sachs, Data Lineage, and Harry Potter Spells

People are still crazy about Python after twenty-five years

10 Questions to Consider when Setting up a Corporate A.I project

What Are Convolution Neural Networks? [ELI5]

The Noonification: Have U Been Pwned? (1/12/2023)

Goldman Sachs, Data Lineage, and Harry Potter Spells

People are still crazy about Python after twenty-five years

10 Questions to Consider when Setting up a Corporate A.I project

Light-Mode

Classic

Newspaper

Dark-Mode

Neon Noir

Minty

HN StartUps