๐ Get Started :Atlas Vector Search Use Cases, Infrastructure, and Product Catalogs
Danny Chan
Posted on August 7, 2024
Atlas Vector Search
Atlas Vector Search Use Cases:
๐ฌ Customer Chatbot
โ Question-Answering (Q-A)
๐ Ecommerce Search
๐ฅ User Recommendations
โ๏ธ Content Generation
๐ Analysis and Summary
Most Common Atlas Vector Search Use Cases:
๐๏ธ Internal Knowledge Bases
๐ Vectorized Documentation
๐ JSON File to an Embedding Model
๐ณ LangChain or LlamaIndex
Atlas Triggers:
๐ Watch for any data changes in a single view
Atlas Vector Search:
๐ Matching documents by similarity search on indexed embeddings data
๐๏ธ Queries can use a vector's metadata (date created) to filter out older content
Atlas Search:
๐ Matching keywords, chunked customer data
๐ค Fuzzy search to correct typos
๐ฎ Autocomplete (suggested search terms)
๐ Index intersection (complex ad-hoc queries)
Infrastructure
Queryable Encryption:
๐ Securing customer data
๐ Encrypt most sensitive data uniquely identifying an individual (e.g., SSN)
Multi-document ACID Transactions:
๐ Integrity of data
Atlas Global Clusters:
๐ Define single or multi-region Zones
๐ Each zone supports write and read operations from geographically local shards
โ๏ธ Configure zones to support global low-latency secondary reads
Atlas Online Archive:
๐๏ธ Data lifecycle management
๐ฝ Automatically send outdated data from active databases into lower-cost cloud object storage
๐พ Keeping data accessible for querying
๐ฏ 9.995% uptime SLA
Distributed Architecture with Elastic Scale:
๐ Dynamically adjust database capacity
๐ Based on application demand (e.g., shopping seasonality, sales promotions)
Product Catalogs
MongoDB Product Catalogs:
๐ฆ Diversity of different products
๐ค Benefit from flexible document data model
Challenges - Keyword Search:
๐ค Without extensive and laborious synonym mapping
๐ฒ e.g., mapping bikes to cycling or sneakers to trainers
Challenges - Recommendations:
๐ง Write complex rules-based engines to get specialized and scarce data
Solution - Product Catalog with Vector Embeddings:
๐ Semantic meaning of products in the catalog
๐ค Understand similarities and relationships between products
Benefits:
๐ Search experience more intelligent & predictive
๐ Track user click-through rates
๐ฐ Sales conversions from search results
More MongoDB Features:
๐ฐ๏ธ Time Series Collections: Ingest and store high-velocity, click-streams
๐ Atlas Charts: Live visualizations of results, continuously tune and optimize business
Editor
Danny Chan, specialty of FSI and Serverless
Kenny Chan, specialty of FSI and Machine Learning
Posted on August 7, 2024
Join Our Newsletter. No Spam, Only the good stuff.
Sign up to receive the latest update from our blog.
Related
August 7, 2024
August 7, 2024