Vector database sharding
Multimodal search
JaguarDB quantization
JaguarDB Vector API
Best Vector databases
JaguarDB in Docker
Setup JaguarDB with tar package
Setup JaguarDB on multiple nodes
Vector index sharing
How zeromove works
Video introduction
|
|
Example: Proximity Search
Combining location proximity search with vector similarity search adds significant value when searching for the most relevant targets that align with users' interests. This versatile approach finds applications in matchmaking, human resources recruitment, discovering automotive sales opportunities, and any service that involves pinpointing specific targets.
The following pseudo-code shows how to find targets meeting both criterias for location proximity and vector similarity:
create store parts (v vector(1024, 'euclidean_fraction_float'), location point(srid:4326), ptime date )
desc = "Vintage Nankai motorcycle road racing suit in good condition, size 58. Has some minor scuffs and scratches but no rips or tears. Very thick leather.")
embeddings = getTextBeddings( desc )
insert into parts values ('embeddings', point(122.42 377.775), '2023-09-18')
desc = "Kuryakyn - Throw-Over Saddlebags, Black in great condition. Designed for riders always on the hustle, Kuryakyn Saddlebags offer additional storage when you need it and leave behind nothing when you don’t. A fully adjustable mounting belt offers throw-over convenience for easy installation and removal without requiring hardware or support brackets. Rugged, purely utilitarian and loaded with ideal storage capabilities for virtually any bike or rider. Includes rain covers and combination lock."
embeddings = getTextBeddings( desc )
insert into parts values ('embeddings', point(122.43 377.776), '2023-09-18')
desc = "I am selling my 2013 Suzuki Burgman 400 abs, it has around 13,530 well maintained miles and is in beautiful condition. Pink slip is in my name, and registration's up to date. All original literature(every single pamphlet), two keys, and factory toolkit are included. $4,000 or best offer. If interested please leave a phone number with a question about the scooter. Suzuki Burgman 400, Burgman, Burgman scooter, Suzuki Burgman 650, Suzuki Burgman 200, scooter, gas scooter, maxi scooter, big scooter, Yamaha Majesty, Honda Silverwing, Honda Forza, Vespa GTS, Yamaha Tmax, Yamaha Xmax"
embeddings = getTextBeddings( desc )
insert into parts values ('embeddings', point(122.42 377.778), '2023-09-18')
desc = "Titled , with Pink slip and off the dmv computer. Complete 1978 rd400 - less wire harness. Great deal on a bike that is in very good condition. Asking $1850(including a couple core 400 bottom ends-( good for cranks,transmissions, heads,cases- damaged) Thanks for looking."
embeddings = getTextBeddings( desc )
insert into parts values ('embeddings', point(122.41 377.773), '2023-09-18')
desc = "I have a 2023 Husqvarna FE501. It has 1hr 30 mins on it with 28 miles not 280000. I have added the following. Power Commander 5, FMF Ti4 Race pipe, Bulletproof Designs Radiator Guards, P3 pipe guard, swing arm Guards, Cycra Wrap Around Hand Guards, BRP Hand Guard Mounts, TM Design Works Skid Plate with link guard, Polisport Clutch Cover, Mother of all Oil filters, Mojo rear disk Guard. It has brand new Dunlap D606 tires installed. The suspension has been sprung and valved for enduro vet by CG suspension. I’m looking to get $11500.00 obo for the bike"
embeddings = getTextBeddings( desc )
insert into parts values ('embeddings', point(122.44 377.779), '2023-09-18')
embeddings = getTextBeddings("Looking for a 5 year old bike less than 80K miles. Budget around $5K ")
select similarity(v, 'embeddings', 'type=euclidean_fraction_float,topk=3')
from parts
where within(location, circle(122.44 377.779 10000)) and ptime between '2023-09-01' and '2023-09-18'
This search aims to identify items that closely match the specified query in the select statement, all while ensuring they meet the location and time window requirements.
|
