Posts Tagged ‘poi’

Location vs. Place vs. POI

With Nokia, Google, Facebook and a whole host of other players recognising the inherent value in the concept of Places and Points Of Interest (POIs), it’s good to see that the World Wide Web Consortium (W3C), the standards body of the Web, is getting involved. On the 30th. September 2010, the W3C Points Of Interest Working Group (POIWG) was launched with a “mission to develop technical specifications for representation of POI information on the Web”. I should pause to make a brief disclaimer here; I’m sitting on the POIWG as part of my day job with Ovi Places at Nokia.

Of course, in order to develop those technical specifications, we need to define what a POI is in the first place. There’s a lot of acronyms flying around (3 in the first paragraph of this post alone) and a lot of conflicting terminology further confusing the matter. Even the most cursory of glances through Web content on this topic shows the terms Place, Location and POI being used interchangably and so as part of the discussion I tried to codify the difference between, and most importantly the inter-relationships between, these three seemingly straightforward terms. The genesis for this post first appeared on the POIWG public mailing list last week (and W3C working groups conduct their business as much as possible in public) but I’ve fleshed it out in a bit more detail here.

Locations vs. Places vs. POIs

So what is a POI? … Wikipedia defines a POI as a Point Of Interest … a “specific point location that someone may find useful or interesting”. But we really need a more subtle and complex definition.

To my mind, a POI is part of a loosely coupled and inter-related geographical terms, comprised of (in generalised order of scope and granularity) Locations, POIs and Places.

A Location is a geographical construct; a physical fixed point on the surface of the Earth. It could also be used to describe a fixed point on the surface of another celestial body but for the purposes of this Working Group, we’ll restrict the scope to terrestrial geographies. A Location is described by a centroid (a longitude and latitude in a widely adopted system, such as WGS-84) and an extent, either a Minimum Bounding Rectangle or a vector set. A Location is temporally persistent, it does not generally change over time.

A POI is a human construct, describing what can be found at a Location. As such a POI typically has a fine level of spatial granularity. A POI has the following attributes …

  1. A name
  2. A current Location (see the commentary below on the loose coupling of POI and Location)
  3. A category and/or type
  4. A unique identifier
  5. A URI
  6. An address
  7. Contact information

A POI has a loose coupling with a Location; in other words, a POI can move. When this occurs, the loose coupling with the previous location is removed and, providing the POI continues to exist, it is then coupled with its new Location. This can happen when the human activity at the POI relocates, such as when your local coffee shop relocates to a new address. It’s still your local coffee shop, it’s now found at a different Location.

A POI has temporal boundaries; it starts when the human activity at that Location commences and ends when human activity ceases, such as when a company or organisation goes out of business.

And then there’s a Place, which is also a human construct and typically has a coarse level of spatial granularity. Places are typically larger scale administrative constructs, either informally or formally defined. Countries, States, Counties, Districts, Neighbourhoods and postal codes or telephone area codes are all Places. Places are also informally or colloquially defined, such as the Home Counties in the United Kingdom and The Bay Area in the United States.

Places have spatial relationships; with parents, children, adjacencies and “contained by” semantics. Places also have the same attribute set as POIs, although with differing interpretations based on scale; for example, the address of a Place or its URI would refer to the address of the administrative or governing body of the Place.

A Place typically contains multiple POIs and can also be coterminous with a POI. In the former case, a Place, such as a city or a neighbourhood, will contain multiple POIs. In the latter case, a Place and a POI will occupy the same position and extent, such as in the case of Yellowstone National Park, which is both a Place and a POI.

As discussions in the POIWG get deeper and deeper into what constitutes a POI and, equally importantly, what doesn’t, it’ll be interesting to see how much of my take on the subject survives.

Written and posted from the BA North Lounge, Heathrow Airport, Terminal 5 (51.474161, -0.484344)

Quantity Or Quality? The Problem Of Junk POIs

In my recent talk to the British Computer Society’s Geospatial Specialist Group, I touched on the “race to own the Place Space“. While the more traditional geographic data providers, such as Navteq and Tele Atlas are working away adding Points Of Interest to their data sets, it’s the smaller, social location startups, that are getting the most attention and media coverage. With their apps running on smartphone hardware, Foursquare, Gowalla and Facebook Places, amongst others, are using crowd sourcing techniques to build a large data set of their own.

For them to do this, the barriers to entry have to be very low. Ask a user for too much information and you’ll substantially reduce the number of Places that get created; and thereby hangs the biggest challenge for these data sets. Both the companies and their users want the Holy Grail of data, quantity and quality. But the lower the barriers to entry, the more quality suffers, unless there’s a dedicated attempt to manage and clean up the resultant data set.

Nowhere is this more evident than in the Foursquare entry for the BCS itself. According to the BCS website, the London HQ of the UK’s Chartered Institude for IT is at The Davidson Building, 5 Southampton Street, London, WC2E 7HA. Now compare that to Foursquare, which lists BCS HQ LONDON as 5 south hompton road, The strand, london, london uk. Complete with interesting use of capitalisation. That’s the first problem.

Foursquare helpfully shows this on a map but evidently uses the provided address information as opposed to any associated geo-coordinate that was gleaned from the onboard GPS on whichever smartphone was used to create this “place“. Google has evidently tried to interpret south hompton road and displays the map at the first entry that Google’s reverse geocoder returns, which is 5 Hampton Road, in Hampton Hill. That’s not Covent Garden. That’s not even Central London. That’s way out in the suburbs of Richmond-upon-Thames. That’s the second problem.

But there’s also more than one entry in Foursquare for the BCS in London which highlights the third problem; large amounts of duplicate Places created by users either unwilling to search too closely for an existing Place or who are trying to subvert the gaming aspect to social location apps in order to gain points or recognition in the community for number of Places created, number of Mayorships gained and so on.

Quantity? Yes. Quality? Sadly no. Foursquare are reliant on their user community to clear up their data and as this example shows, that’s not always an effective strategy. As an industry we may be building a massive Place based view of the world but we’ve a long way to go before we can rely on data produced in this manner.

A geographic nod of the hat must go to Harry Wood for spotting this classic example of a “junk POI“; I’m not singling Foursquare out for any particular opprobrium here by the way, all of the social location data sets have their own howlers, as do the commercial POI data sets, ready and waiting for people to stumble across.

Written and posted from the Nokia gate5 office in Berlin (52.53105, 13.38521)