Open Address presentation to ODI Devon on 18 March, includes company overview and details on inference (aka multiplying addresses) and smart address input
3. Overview
In the UK addresses are currently a
closed asset owned by private and public
sector institutions.
Open Addresses make address data
available for everyone to use for free,
build a community to maintain it, and
create paid-for services to fund it.
Open Addresses is wholly owned by the
not-for-profit Open Data Institute.
Maintain
Publish
Gather
https://openaddressesuk.org
@peterkwells
4. Benefits
Social and economic value Speed and quality
We estimate that opening up UK
addresses will create yearly social
and economic value of >£110m as
more people will use them
We will reduce the time it takes
for one of the UK’s 130,000 new
addresses per year to get published from
months to days
New businesses and services Efficiency
Resolvable identifiers will allow
new businesses and services
to be created through linked data
We will provide address services that are
cheaper and simpler for service providers
to use, integrate and maintain
https://openaddressesuk.org
@peterkwells
5. How people will use the services
● To enter free-format addresses
● To correct bad data
● To locate addresses
● To link data together
● To create new services
https://openaddressesuk.org
@peterkwells
6. How we gather data
Alpha Beta Live
Bulk open datasets, crowdsourcing
Large, closed datasets?
Statistical & geospatial inference
Collaborative maintenance
https://openaddressesuk.org
@peterkwells
7. Collaborative maintenance
The more users on the platform the higher the quality of the data
Service providers and users
automatically produce information about
new and existing addresses as they use
the services.
Our algorithms weight this data to
improve the data and service.
AddressBase
PAF
https://openaddressesuk.org
@peterkwells
8. How we will grow the user base
With sponsorship we can grow the user base faster by extending free
trialist services
Alpha Beta Live
Open/geo community
Social comms
Commercial trialists
Mainstream media
Public sector trialists
Technology media
Paying customers
https://openaddressesuk.org
@peterkwells
11. Data: changing needs
As we build an authoritative, definitive dataset we expected data needs to change over time
Existing
open
datasets and
inference to
tackle the
bulk of the
challenge
“80/20” rule
Ongoing,
collaborative
maintenance
Targetted
work. Low-
volume
records to fill
existing gaps
in available
datasets
NB: dates are “just for fun”
https://openaddressesuk.org
@peterkwells
12. Data: changing needs
But IPR got in the way of existing open datasets
Many are not really open
Ask me about digital cholera :-(
We had to adapt our plans
We’re both growing data and growing a network….
23. Fogralea, Sound, ZE1 0SE
● Map: sourced from OS Street
View
● Yellow dots: street mid lines (from
OS Locator)
● Blue lines: street geometry (from
OS VectorMap District combined
with OS Meridian 2 and OS
Locator)
● Red dots: postcode centroids
(from ONS’ Postcode Directory).
● Red hatched area: street buffer
zone 25m either side of the street
mid line (calculated)
24. Meadow Lane, Chester, CH4 7BH
● Map: sourced from OS Street
View
● Yellow dots: street mid lines (from
OS Locator)
● Blue lines: street geometry (from
OS VectorMap District combined
with OS Meridian 2 and OS
Locator)
● Black lines: Land Registry’s
INSPIRE Index Polygons
● Red dots: postcode centroids
(from ONS’ Postcode Directory).
● Green rectangles: street
bounding boxes (from OS
Locator)
● Red hatched area: street buffer
zone 25m either side of the street
mid line (calculated)
Note: the extra features are available thanks to the Land
Registry’s open data offering for England and Wales.
26. Collaborative maintenance
The more users on the platform the higher the quality
Service providers and users
automatically produce information about
new and existing addresses as they use
the services.
Our algorithms automagically weight
this information to improve the service.
AddressBase
PAF
https://openaddressesuk.org
@peterkwells
27. The Open Addresses platform
Our aim is to build a sustainable model like this. The value is in an authoritative and definitive dataset
which is collaboratively maintained.
Authoritative and definitive UK address list
Bulk APIs URLs
Service Providers
Services
- Websites,
Users
Value
Revenueforsustainability
28. The (lack of a) network effect
We have one million high-confidence addresses [*] so it looks more like this. Despite the collaborative
maintenance we lack authoritative and hence little chance of becoming definitive
Authoritative and definitive UK address list
Bulk APIs URLs
Few Service Providers
Services
- Websites,
Few Users
Revenueforsustainability
Value
29. Building Blocks
If we drill into the platform we see something interesting. The building blocks for addresses are
creating value in our address list. Can we build a value proposition from this?
APIsBulk URLs
Value
Building Blocks
- towns
- postcodes
- streets
- and the links between them...
Authoritative and definitive
UK address list
Revenueforsustainability
30. Smart address input
One of the needs that building blocks can support
is improving address entry on websites
This can meet user needs, provide value to service
providers and provide clean addresses for the
address list. Win! Win! Win!
We are building a proof-of-concept…..
31. It will give people less of this
Giff Gaff on desktop
Hungry House on mobile
The Jigsaw Puzzle Club
Lots of fields
Exceptions
BIG dropdown lists
Burn your select tags:
https://www.youtube.com/watch
?v=CUkMCQR4TpY
32. And more of this
Citymapper
High digital confidence/ skills
users like suggestions.
We have a PoC...
Low digital confidence/skills users like to
just type. Do the hard work….
http://sorting-office.openaddressesuk.org
“is this what you meant?”
(Dev/Alpha language at the mo…)
33. Smart address input
It can be delivered with a small address list
It can handle “digital cholera”
It helps users
It helps services providers
It helps build a public good
36. How you can help
As a service provider As a developer / open data advocate / geospatial
expert
Think of how you would use our services and let us know Add more ideas or add “ready to load” datasets:
https://trello.com/b/UXqgc1wy/datasets
Become a Beta trialist for the APIs Load some data:
http://turbot.openaddressesuk.org/
Let us know if you have data you are willing to contribute Visualise and explore the current bulk download:
https://alpha.openaddressesuk.org/data
Let us know if you would be willing to sponsor us whilst we grow Comment on our algorithms:
https://alpha.openaddressesuk.org/blog/2015/02/12/inference
https://alpha.openaddressesuk.org/blog/2015/02/20/confidence
mailto:info@openaddressesuk.org Try our APIs:
https://alpha.openaddressesuk.org/about/apidocu
Contribute some code: https://github.com/OpenAddressesUK
Join the forums: https://github.com/OpenAddressesUK/forum
38. Open Addresses is building an open address
database for the UK.
It will be open and available to all, helping people,
companies, non-profits and government work
together to create a better service for everyone.
Offices
Where?
Editor's Notes
Some of below still needs formatting….
Social and economic value
Reducing the cost of buying and using addresses will lead to the creation of new businesses and new services with increased social and economic value
If the UK realises similar benefits to that seen by an open addresses initiative in Denmark, the UK economy will see new yearly benefits of > £110m [1] with a significant increase in the market size of address users
We are building and maintaining our data using modern, highly automated methods that significantly reduce maintenance costs over legacy models
We are importing all data using a model that makes the published address safe to use by anyone for any purpose [2]
Speed and quality
It can take months for new addresses, particularly those with new postcodes, to be made available [3]
When people move into a new house they need to buy insurance, a pizza, or register to vote but because of this issue they often can’t. This leads to lost revenues and potentially more significant problems due to service providers not being able to find an address [4]
Over the last 10 years the UK has, on average, seen.130,000 new homes every year [5]
Our collaborative maintenance and confidence model will allow any person or organisation to create or modify an address. Our users will be provided with information to allow them to judge how accurate an individual address record is likely to be.
Digital/online services
57% of users use mobile to access the Internet [4], up from 49% in 2013
Desktop-optimised websites either offer multiple fields (house/road/town/postcode) or prompt for a postcode before offering the user a drop-down list
Mobile usage is growing [1] yet the same designs predominate despite the even worse user experience on smaller screens
Location and map services [2] have developed new design patterns, such as free-format text entry, but rely on historic search data/personal data that many sites lack
Open Addresses will open up APIs that will bring many of these capabilities to more services without the requirement to capture historic search data/personal data
New businesses and services
Denmark saw a 1000% increase in orgs using address daa
Multiple departments/applications within the same organisation frequently use different formats leading to an inability to link data and high costs due to manual intervention
Multiple organisations in the same industry often use different address sets leading to an inability to link data and service failures [3]
This problem can be partially solved by the use of proprietary datasets containing a UPRN (unique property reference number) issued by public sector bodies, but these datasets trigger issues with cost and data quality
Open Addresses will support multiple presentation formats from a common data model whilst issuing each address with a unique persistent URL that can be freely used by anyone
We expect this model to encourage new services to be created, for example a service to provide logistics firms with common delivery instructions for an address
The UK government’s main web presence, GOV.UK sees continued growth in mobile usage, reaching as high as 31% of all visits: https://www.gov.uk/performance/site-activity
For example Google Maps, Apple Maps, Swarm or Citymapper
Two simple examples. First, the telecoms industry which needs to exchange addresses in common formats to perform directory enquiries or identify premises that have requested a broadband installations; second the public sector that needs to identify and manage service problems across multiple delivery organisations.
http://media.ofcom.org.uk/facts/
“In 2010 it is estimated that the social benefits from the agreement will be about EUR 14 million, while costs will total about 0.2 million - See more at: http://www.epsiplatform.eu/content/value-danish-address-data#sthash.66jSkCtU.dpuf”. £110m is a pro rata figure using population size as a proxy for address volumes. Denmark saw a 1000% increase in the number of organisations using address data.
https://alpha.openaddressesuk.org/blog/2015/01/26/making-address-data-safe
The Royal Mail address advisory board states up to 13 weeks for just the Royal Mail element of the process: http://www.pafboard.org.uk/documents/PAF(15)2%20Strategy%20for%20PAF%202014.pdf. In addition to the Royal Mail activities there is work to do in local authorities, Geoplace and possibly Ordnance Survey. Following those activities an address is published and the user then has to wait for service organisations to choose to update their lists.
https://openaddressesuk.org/blog/2015/02/09/living-breathing-problem
https://www.gov.uk/government/statistical-data-sets
Note that many open datasets cannot be used as they are understood to contain third party IP, e.g. LandReg. This highlights a problem with current open data and how tangled the addressing picture is.
Large closed datasets ::: BT phonebooks!
Mathematical inference: https://alpha.openaddressesuk.org/blog/2015/02/12/inference
Collaborative maintenance: next page
When we have sufficient users we can incentivise feedback through pricing mechanisms.
Feedback mechanisms via APIs
searching for an address
this address is right
this address is wrong
this is a new address
Confidence is how we value feedback: https://alpha.openaddressesuk.org/blog/2015/02/20/confidence
Conversations with trialists underway…..
We can’t use Google Maps to check the shape of the street and the location of the buildings, because it’s foggy…
… no, because its terms don’t allow us to!
When we have sufficient users we can incentivise feedback through pricing mechanisms.
Feedback mechanisms via APIs
searching for an address
this address is right
this address is wrong
this is a new address
Confidence is how we value feedback: https://alpha.openaddressesuk.org/blog/2015/02/20/confidence
We need money to flow from top-to-bottom to keep the service running. Yes, there’s other revenue streams like sponsorship/donations, ProfSvc but this is what a sustainable platform in the private sector would look like
Value flows in two directions. This is the model for collaborative maintenance.
Some services can be provided by Open Addresses UK, need to be careful when competing with own customers.
[*] inference adds more but low confidence
Give us your data! But we can’t give much in return…..
This is a problem unless we can find a copyright-correct way of bringing data in. Working on that but there’s a chance we might fail….
Note that value is flowing both ways. The address list gives value to the building blocks by helping us learn which links are true (for a given level of true), i.e. postcode SE5 is in London, postcode FY8 could have towns Lytham or St. Annes
The Library of Babel: http://en.wikipedia.org/wiki/The_Library_of_Babel is a literary analogy of what happens if you have blocks, but not links