VectorSearch-Cities ObjectBox C++ Example

This example demonstrates vector search in ObjectBox using capital cities. You can search cities according to their geolocations (latitude and longitude), e.g. find the closest cities to a given city or a given geolocation. It uses a simple command-line interface; see below for some example queries.

2D vs. high-dimensional vector search

We chose this example as an intuitive way to demonstrate ObjectBox vector search; e.g. cities with 2D location vectors are relatively easy to understand and work with. While this is a perfectly valid use case, ObjectBox vector search is primarily built for high-dimensional vector data. Thus, ObjectBox vector search is typically used for embeddings produced by AI models, which have hundreds or thousands of dimensions (e.g. there's no limit on the number of dimensions).

Build and run

You need CMake 3.14+ and a C++14 compiler to build this example. Run ./build.sh run is the simplest option; other options are described in the build and run section of the parent directory.

Usage examples

Once you start the example, you will see something like this:

Welcome to the ObjectBox vectorsearch-cities app example
Available commands are:
    import <filepath>          Import CSV data (try cities.csv)
    ls [<prefix>]              List cities (with common <prefix> if set)
    name <city>[,<n>]          Search <n> cities to nearest to the given <city> name/prefix
                               (<n> defaults to 5; try `name Berlin` or `name berl`)
    geo <lat>,<long>[,<n>]     Search <n> cities nearest to the given geo location
                               (<n> defaults to 5; try `geo 50,10`)
    add <city>,<lat>,<long>    add location
    removeAll                  remove all existing data
    exit                       close the program
    help                       display this help

Now, you can interact with it by typing in commands.

List all entries with ls:

ls
[..]
207  Willemstad          12.11     -68.93   
208  Windhoek            -22.57    17.08    
209  Yamoussoukro        6.83      -5.29    
210  Yaoundé            3.85      11.50    
211  Yaren               -0.55     166.92   
212  Yerevan             40.19     44.52    

Search nearest neighbors to Berlin with name Berlin, 10 (note that "Score" is the distance in kilometers):

name Berlin, 10
 ID  Name                Location            Score     
 28  Berlin              52.52     13.40      0.00
147  Prague              50.08     14.44     281.13
 49  Copenhagen          55.68     12.57     355.15
203  Warsaw              52.23     21.01     517.17
200  Vienna              48.21     16.37     523.55
 34  Bratislava          48.15     17.11     552.40
  6  Amsterdam           52.37     4.89      576.75
 94  Luxembourg City     49.61     6.13      601.97
 37  Brussels            50.85     4.35      650.65
196  Vaduz               47.14     9.52      659.56

Search nearest neighbors to Area 51 via latitude and longitude with geo 37.23, -115.80 (note that "Score" is the distance in kilometers):

geo 37.23, -115.80
 ID  Name                Location            Score     
107  Mexico City         19.43     -99.13    2555.95
204  Washington D.C.     38.91     -77.04    3372.90
129  Ottawa              45.42     -75.70    3431.69
 27  Belmopan            17.25     -88.76    3452.91
 64  Guatemala City      14.63     -90.51    3542.07