Find closest match for home address - mongodb

Question

Let say one of my property object in the database looks like:
{
id: ObjectID(454379857349857349545345),
property_name: 'building 9',
address_street_1: 'mark street 2',
address_street_2: 'mark avenue',
address_city: 'o town',
address_zip: '687',
state: 'o state'
}
I need to check if there is already a property of the same/similar address is already created before creating a new one. If let's say the new data has the address of:
property name: 9,
street 1: mark street 2,
street 2: mark ave,
city: O-town,
zip: 687,
state: O-state

Answer 1

I would personally not try to be matching on things like "building 9" being the same as "9" or any other combination. It's just too broad and prone to error in the logic, and far to much logic to code.
If you stored the "geolocation" for the property, then you could actually do a direct comparison that you are not adding something with basically that same location data or "near" that location by too small a distance. You can get a distance accurate to 1 meter, so anything generally within a few square meters would be a "no" as it would be the same place, or at least up to scrutiny.
So if your data in fact looked like:
{
_id: ObjectID(454379857349857349545345),
property_name: 'building 9',
address_street_1: 'mark street 2',
address_street_2: 'mark avenue',
address_city: 'o town',
address_zip: '687',
state: 'o state',
"location": {
"type": "Point",
"coordinates": [150.9482,-33.7764991]
}
}
And you then wanted to check a new listing by it's location data, then you would basically be doing a $nearSphere query with $minDistance to filter out anything too close:
db.collection.find({
"location": {
"$nearSphere": {
"$geometry": {
"type": "Point",
"coordinates": [150.9482,-33.7764991]
},
"$minDistance": 20
}
}
})
Where there is a minimum distance of 20 meters and you generally are not expecting to find anything in the results. But if you do then it is a quick check of the returned list to see if this "looks like" the same address or not, and that can be handled by a "human".
So with something concrete such as a GPS reading from the location and keeping it stored, then checking adresses that are the same becomes a simple task.
Look at Geospatial Queries and indexes for more information.
Addendum
Just thought this to actually be a useful search result, so here is the basic subtext of how you even get to this point.
For such a system as is "generally described" in the question, you typically have an "agent" who is responsible for the property listing (sic), and that it is a very reasonable presumption that "said agent" is going to visit that property of interest in the course of establishing new business.
Therefore two basic points come to mind:
Who on this planet ( in a reasonable employed scenario ) is not carrying around a mobile phone equipped with the capability to work out their current location? Even cell tower signal strength triangulation should get this mostly right to within a reasonable meter radius, but generally these "little computers" all come with GPS capability, and also the ability to interact with an application that records their current location.
There may be exceptions to this, or just the "non tech savvy super salesperson" who at the very least can either "point at a map" ( they actually found the place somehow afterall ) or work it out themselves in the office by selecting a map location from an interface.
Considering either of those two options, then a user ( or helper ) would need to be completely inept to not be able to select the location of the property from a presented map in some way. Hey, "Google" ( no branding intended ) will get it right most of the time just based on the "text" of the typed address. But we are not building "Google" ( brand endorsement not intended ) here, but just an App that is going to do the job.
At the end of the day, if people are not really capable of "pointing at a map", or otherwise using the GPS they carry in their pocket, then let them look it up manually. And just list it as a "design constraint", until the $20K bill gets paid.