Thursday 29 March 2012

Review of On Location event

ISKO UK and the British Computer Society co-hosted an event all about location data.

The first speaker was Mike Sanderson of 1Spatial, who described using geo-spatial data as helping to power a European knowledge economy. He spoke of the need for auditing and trust of data sources, particularly ways of mitigating poor quality or untrustworthy data. At 1spatial, they do this by comparing as many data sources as they can to establish confidence in the geodata they use. Confidence levels can then be associated with risk, and levels of acceptable risk agreed.

Alex Coley of INSPIRE talked about the UK Location Strategy, which is aiming to make data more interoperable, to encourage sharing, and to improve quality of location knowledge. The Strategy is intended to promote re-use of public sector data, and is based on best practice in linking and sharing to support transparency and accessibility. Historically there has been a lot of isolated working in silos, and the aim is to try to bring all such data together and make it sharable. This should help organisations to cut costs in technology support and reduce unnecessarily duplicated working. Although some organisations need to have very specialised data, there remains much that is common. Location data is frequently present in all sorts of data sets, and can be re-used and repurposed, for example to help understand environmental issues. Location data can be a key to powering interesting mashups - for example someone could link train timetables with weather information, so train companies could offer day trips to resorts most likely to be sunny that day.

The Location Strategy's standardisation of location data is effectively a Linked Data approach, but so far little work has been done to map different location data sets.

Data that is not current is generally less useful than data that is maintained and kept up to date, so data sets that include information about their context and purpose are more useful.

Jo Walsh of EDiNA showcased their map tools. EDiNA is trying to help JISC predict search needs and provide better search services. They are trying to take a Linked Data approach, but there is a need for core common vocabularies. EDiNA runs various projects to create tools to help open up data. Unlock is a text mining tool that helps pick out geo location data from unstructured text. It could be used to add location data as part of digital humanities projects.

One aspect of Linked Data that is often overlooked is that it can "future proof" information resources. If a project, or department, is closed down, its classification schemes and data sets can become unusable, but if stable URIs have been added to classification schemes there is more chance that people in the future will be able to use them.

Matt Bull of the Health Protection Agency explained how geospatial data is useful for health protection, such as tracking infectious diseases, or environmental hazard tracking. Epidemiology. Data is inherently social. Diseases are often linked to the environment - radon gas, social deprivations - and clinics, pharmacies, etc. have locations. This can be used to investigate treatment seeking behaviour as well as patterns of infection, in order to plan resourcing. For example, people often don't use their nearest clinic especially in cities, to seek treatment for sexually transmitted diseases. Such behaviour makes interpretation of data tricky.

Geospatial data is also useful for emergency and disaster planning and monitoring the effects of climate change on the prevalence of infectious diseases.

Stefan Carlyle, of the Environment Agency (EA), talked about the use of geospatial data is in incident management. One example was using geospatial information to model risk of failure of dam and plan an evacuation of the relevant area. Now it is a quick and easy operation that would not have been possible in past without huge effort. Risk assessments of flood defences can now be based on geospatial data and this can help prioritise asset management - e.g. management of flood defences.

Implementing semantic interoperability is a key aim, as is promoting good quality data, and this includes teaching staff to be good "data custodians". Provenance is key to understanding the trustworthiness of data sets.

The EA is focussing on improving semantic interoperability by prioritising key data sets and standardising and linking those, rather than trying to do everything all at once. Transparency is another important aim of the EA's Linked and Open Data strategy and they provide search and browse tools to help people navigate their data sets. Big data and personal data are both becoming increasingly important, with projects to collect "crowd sourced" data providing useful information about local environments.

The EA estimates that its Linked Open Data approach produces about £5 million per year in benefits from reduced duplication of work and other efficiencies such as unifying regional data silos, and from sale of data to commercial organisations. The EA believes its location data will be at the heart of making it an exemplar of pragmatic approach to open Data and transparency.

Carsten Ronsdorf from the Ordnance Survey described various location data standards, how they interact, and how they are used. BS7666 specifies that data quality should be included, so the accuracy of geospatial data is declared. Two key concepts for the OS are the Basic Land and Property Unit and the Unique Property Reference Number. Address data is heavily standardised to provide integration and facilitate practical use.

Nick Turner, also of the OS, then talked about the National Land and Property Gazetteer. The OS was instructed by the government to take over UK address data management because a number of public sector organisations were trying to maintain separate address databases. The OS formed a consortium with them and formed AddressBase.

AddressBase has three levels - basic postal addresses, AddressBase plus which includes addresses of buildings such as churches, temples, and mosques, and other data, and AddressBase premium, which includes details of buildings that no longer exist and buildings that are planned.

AddressBase is widely used as it allows organisations to refer to AddressBase to verify and update or to extract other address information that they need when they need it, rather than having to manage it all by themselves.