Points of interest are all around us, but what are they? And how are they created?
A point of interest (POI) can be considered as any location that someone may be interested in. This could be anything from a bin to a mountain. All that’s required is a name and a location, but often they’re tagged with a description and category too so that they can be searched, analysed and plotted with ease.
A great place to start when looking for POIs is OpenStreetMaps (OSM). There are millions of community mapped POIs, covering the entire planet and each tagged with tonnes of useful information.
In the random portion of OSM map below, a few points of interest have been circled. Here, the points of interest take on the form of a Park, Hospital, Bridge, Airport, and Carpark.
.png)
OSM POIs are tagged with standardized categories where possible to make plotting and analysis easy. The POIs in the map above have been given tags like amenity: hospital so when plotted, the correct icon can be used to represent the POI. Websites like taginfo and the OpenStreetMaps wiki provide plenty of information on what tags are present and what the standardized names are for a variety of POIs.
Sticking with the hospital theme, websites like OpenStreetBrowser enable you to plot the specific POIs you’re interested in from OSM on a map. Here’s more of London with health-based amenities explicitly plotted.
.png)
Outside of the OSM sphere, there are hundreds of other datasets, APIs, and services for accessing POI data. ESRI, Google, Microsoft, TomTom, and Ordinance Survey (OS) all provide such products. So, if you’re on the lookout for some POI data, how do you know what’s good, and what’s not?
When looking at POIs, there are three key factors you’ll want to assess.
Everyday there are new businesses opening, some businesses changing hands, and others closing. To provide the end user with accurate information POI data needs to be kept up to date. It’s a great idea to check the update frequency of the your POI data, as well as the average age of the POIs in each region. Otherwise, you might be sending someone on a wild goose chase.
Ensuring good coverage in a region is vital. Without coverage, it’s likely users of your maps, search, analysis, or otherwise, will be unhappy as you don’t have what they’re looking for! Unfortunately, this is the hardest part of assessing the quality of a POI dataset. How do you know when you have 90% coverage, or 99% coverage? You can’t. This is where user feedback and contributions come in. Without it, it’s simply impossible to have every POI or know what POIs are accurate and what POIs aren’t.
Ensuring POIs are consistent is a big task. You need consistency not only in location, descriptions, and names, but also in the categories. The OpenStreetMap foundation have regular proposals to update and improve different tags to ensure consistency throughout the entire database. A tricky task when the database contains millions of legacy data points with millions of users relying on it. Make sure you start off with a sensible classification scheme!
Location accuracy is important too. Ensuring locations are accurate is essential as this data is relied upon to get people from point A to POI B. A tiny change in the coordinates of a location could mean it gets placed on an adjacent street or building leaving anyone trying to navigate there lost.
Now we know how to assess the quality of POI data, let’s dive into how it’s collected so you can make the most informed choice when looking for POI data.
It’s often of interest for business owners to publicly post the location, name, description and opening hours of their business online to ensure that it’s as easy as possible to find. This may take the form of a Facebook page, a listing on Google Maps, or adding it to OpenStreetMaps - either way it eventually trickles down into POI databases.
OSM users often spend large amounts of time mapping their local area. Their knowledge of the area is good and, with the aid of online tools which help overlay maps and satellite images, the positions of trees, benches, bins, bus-stops, parks, lampposts, and more can be directly added to the OSM database.
Rather than directly mapping POIs, a variety of open source projects collate different datasets into one complete dataset. For example, All the Places asks users to generate web spiders to scrape the location, name, and opening times of various different stores and companies across the world from their websites. The scripts are then run every week and the data updated.
Private companies such as Google, TomTom, and Microsoft have been building up data for decades. The Google street view car has imagery of over 10 million miles worth of road, meaning there are hundreds of millions of images full of road signs, bus stops, tram lines, and anything else you’d find on the road! In addition, these companies also utilise vast amounts of satellite images to build up a picture of road networks, buildings, and anything else that may be of interest for navigation.
Hopefully you now know a little more about points of interest than you first did when you clicked on this blog! With so much data out there, we’d love to hear about any cool projects you’re working on involving point of interest datasets.
Sign up at Naurt for product updates, and stay in the loop!
