Steve Calder: Business Benefits of GIS: An ROI Approach
Richard Smith: Addressing the Problems of Addressing at British Transport Police
1. Addressing the Problems of
Addressing
at
British Transport Police
Richard R. Smith,
Force Information Manager, British Transport Police
Bob Chell,
Principal Consultant, 1Spatial Group Ltd
2. • National Uniformed Police Force formed in
1826 (3yrs before the Metropolitan Police
Service)
• 10,000 miles of track
• 3,000 railway stations and depots
– National Rail network across England,
Scotland and Wales
• London Underground system
• Docklands Light Railway
• Glasgow Subway
• Midland Metro tram system
• Croydon Tramlink
• International services operated by Eurostar
Who are BTP?
3. Who are BTP?
- Police the journeys of six
million passengers
- 400,000 tonnes of freight
- Over 10,000 miles of track
We believe traveling is about
more than just getting there.
It’s about ensuring safety and
security all the way.
4. UK Location Strategy
Place Matters. Everything happens somewhere. If we
can get a better understanding on this we can make
better use of resources, improve planning and advance
our management of risk.
There is too much data duplication, too little reuse, too
few linkages across datasets to support policy
implementation.
This is particularly true for the emergency services,
where inaccurate data can result in lives being put at
risk.
14. The Problems of Addressing
A root cause of these inaccuracies is the multiple
sources of event data and often conflicting address
databases in use, such as NLPG, AL2 and PAF.
There is an initiative (NESG) to produce a reliable,
common address gazetteer for the emergency services,
which will overcome many of these problems.
However, there is an immediate need for individual
forces to maintain their own address database or
gazetteer now. This is for both incident response and to
provide accurate mapping in support of intelligence
generation, resource planning and many other activities.
15. Addressing the Problem
British Transport Police (BTP) has recently
embarked on a programme to automate the audit and
repair of their incident database in relation to the NLPG
and other address files.
BTP have been able quantify the quality of data held
within their Gazetteer, which is core to providing location
information to any Officer responding to an Incident on
the Railway.
After the audit, the same technology will be used to
provide ongoing validation and ensure data integrity and
reliability.
16. The Impact on the Business
“Right Information, Right Place, Right Time”
• Assist in meeting National Targets
NATIONAL TARGETS FOR ALL AREAS
OBJECTIVE TARGET
FATALITY MANAGEMENT To conclude police activity which disrupts
train movement within an average of 90
minutes from receiving a report of a fatal
incident.
This target excludes Major Incidents and incidents classed as Suspicious, RTC Level Crossing,
Unexplained, Sudden Deaths and Work related deaths.
17. Master Data Management
BTP have achieved this by generating a
baseline of information based on the NLPG
and NSG against their Location Gazetteer.
A rule-based approach has been taken to
evaluate the data and build this baseline. In
effect, they have generated a master index
of their Location Gazetteer.
This index provides a complete and
consistently assembled view of what is
happening and where.
24. ATTRIBUTE VALUE
LOCATION THORNE NORTH RAILWAY STATION
TYPE RAILSTN
STREET FIELDSIDE
POST_TOWN DONCASTER
POST_CODE DN8 4HZ
ATTRIBUTE VALUE
LOCATION THORNE NORTH RAILSTN
STREET
POST_TOWN DONCASTER
POST_CODE
Matching Process to Populate the Index
25. BTP Impact Assessment
The BTP Locations are first analysed to understand
the quality (completeness and logical consistency) of
the data. It allows us to make a baseline assessment of
the information.
Business rules check different address
characteristics of the data, focussing on the address-
related elements such as Street Name, Postcode
and Post Town.
26. Index Population - Rules
These checks are applied to themes of BTP location
types. Different groups will contain subtly different
characteristics.
Grouping the data into themes also reduces the number of
rules that BTP have to construct. This makes it easier to
manage the rule base.
Stations (Railway, Disused, …)
Tier 1 (Control Rooms, Metro, …)
Tier 2 (Bridges and Crossings, …)
Tier 3 (Signal Boxes, Sidings, …)
27. Populating the Index using Rules
If
NOT IN THE INDEX
or
PART OF THEME
Then perform the
necessary
conditional checks
If the checks are
true then populate
the index
31. Summary - ALL Total Percentage
Not Matched 1166 15%
Matched 7216 94%
Total 7691
Index Overview with Confidence
32. BTP Content, Update
Less than 10% of the data matches the NLPG or NSG
But we have matched 60-85% of the data, with confidence
BTP
NLPG
33. BTP Content, Append
Less than 20% of the data even contains a Street Name…
…similar characteristics for Postcode and Post Town
???
But we have matched 60-85% of the data, with confidence
BTP
NLPG
34. Advantages
• Leveraging Intelligence
• Information Re-Use
• Accuracy (fit for purpose)
• Rules-based Approach
• Data Centric Intelligence
• Maintaining History
35. Benefits
• Reliable Data – Saving Time, Saving Lives
• Location
• Correct Allocation of Resources
• Correct Incident Reporting
36.
37. Benefits
• Reliable Data – Saving Time, Saving Lives
• Location
• Correct Allocation of Resources
• Correct Incident Reporting
• Addressing the Management of Police
Information (MoPI)
Editor's Notes
Rules engines makes it easier to program allowing us to move from imperative models, to lists of production rules.
Four generic index population rules are applied in two sessions, using NLPG then NSG.
The checks are applied to the themes of BTP location types. This is because different groups will contain subtly different characteristics.
Separating the rules into separate workflows keeps the rules within a narrow context, making the impact programme flow easier to manage. It also allows BTP to keep extending the index, dataset by dataset.
The system runs through the rules, picks the ones for which the condition is true, and then evaluates the corresponding actions, in this case populating the index.
The interaction of the implicit rules can often be quite complex, particularly when the actions of rules impact the resulting conditions of other rules.
Chaining rule-based tasks in different orders can lead to different results. Support for re-ordering is essential in order to systematically discover the flow that populates the index with optimal confidence.
The addressing problem, plus many other problems, fit this production rule computational model.
Leveraging Intelligence
The index enables users to understand, with confidence, that the same place can described in different ways in different data sources which means they can truly understand the nature of place.
Information Re-Use
The index enables the user to re-use different data. This means users can perform data level comparisons to define baseline and quantify data quality in data sources.
Accuracy (fit for purpose)
Use the index of information by updating and appending information from one source to enable yourself to automatically improve the content and accuracy of data sources.
Rules-based Approach
Managing production rules, not data models, enables the business people to use specialist knowledge. They can control and describe the rules without the expense and dependency on specialist skills or developers.
Data Centric Intelligence
Use the index information to define database views of the information, which you can control and administer once and deploy across the entire enterprise.
Maintaining History
Use changes in the index to control information archiving. This enables users to maintain historical views of information, as well as keep an auditable and versioned data source.