Telstar

Erewhon

Erewhon is a project based at

OxfordUniversity looking at developing geolocation services and to improve mobile access to resources – being presented today at an Open University Library Seminar by Sebastian Rahtz and Tim Fernando. I hope I’ll have a chance to do a follow up blog post soon on how I think this work could be relevant to TELSTAR – but for now, you’ll just have to enjoy these very brief notes from the session.

Erewhon is funded (as is TELSTAR) under the JISC Institutional Innovation programme. The aim is to:

• Develop geolocation services
• Improve mobile access to resources

Erewho is predicated on a number of assumptions:

• Free mapping
  • Mapping things using geographic metadata is become ubiquitous
• Institutional geodata will be available
  • Locations of major buildings etc.
• Smart phones will become increasingly common (towards saturation)
• Increasingly there will be usefully personalised data in systems
• Users can supply their location automatically
  • GPS, CellTower ID, Wifi ID, FireEagle etc.

Deliverables are:

• Implementation of a geolocation data model, syndication interface and API for organizational resources
• Provision of demonstration location-aware applications
• Adaptation of selected Sakai tools for use on relevant mobile devices
• Guidelines for others to implement similar tools

The project has a blog at http://oxforderewhon.wordpress.com/

Erewhon is not aiming to write the applications that make use of the geolocation data, but ensure there is a stable platform for such applications to be built on.

What are the issues with current approaches (e.g. Google maps)?

• Google location of political units is patchy and not always accurate
• Fixing the Google data when you find a mistake is cumbersome
• They do not model to the level of the entrance or room
• Their apparent knowledge of university structure is simply based on keyword searching

We need systems that can handle:

• Time based data
• Rich relationships
• More entities
• Extendable

RDF was the obvious way to go.

Decided to implement with Gaboto – an open source generic RDF storage engine that allows for automatic mapping from RDF to Java objects and is able to cope with time in RDF.

Result is 2nd generation of Oxpoints – it covers colleges, departments, libraries, museums and carparks, but also anything else one may encounter.

The types of objects that are recorded are things like:

• Building
• Division
• DrainCover
• Entrance
• Faculty
• Group
• Image (image related to the thing?)
• Room
• ServiceDepartment
• Site
• WAP
• Website

Have collecting information on about 1000 ‘political entities’- e.g. libraries etc.
Data collected from existing databases, web pages, walk-rounds, Google Earth, Estates department.
You can query Oxpoints for a KML results set – which can e.g. be pushed to GMaps.
http://maps.google.co.uk/?q=http://m.ox.ac.uk/oxpoints/colleges.kml
You can explore the dataset at http://m.ox.ac.uk/cgi-bin/oxshow.pl (no SPARQL endpoint at the moment though)

Mobile technology:

• About supporting more than just ‘reading email’
• Smartphone market eating into traditional mobile phone market
• Essentially – all phones will become ‘smart’

What are institutions doing with mobile technology:

• http://m.mit.edu
• iStanford app for iPhone

Oxford now has http://m.ox.ac.uk

This gives access to a selection of custom services including some which are firsts for any University:

• Let me listen to some podcasts by Humanities
• What’s my College Tutor’s name
• Where’s my nearest postbox and what time is it collected at?
• Which is the nearest library with book X
• How do I cycle from A to B?

Fast Development Process – Agile programming methodology with Continuous Integration (using Hudson). Site is built using Python and django framework.

An example of the type of application they want to be able to provide is real time bus information:

• DoT provided NaPTAN (National Public Transport Access Node) database under licence (free under educational license) – lists every bus stop, platform, port etc. with geocode
• Removed irrelevant data (bus stops that no longer exist, or aren’t currently used) – not a trivial task

Workflow is:

• Mobile user browses to m.ox.ac.uk
• Django gets items (e.g. bus stops) within a certain radius
• Use Open Street map to get some modified map tiles
• Then request bus data from oxontime.com
• Adds details to map and returns to browser

This is currently in private beta

Some other services that they are planning, or interested in are:

• Access webcam feeds from across Oxford
• Access to Results schedule (when different sets of results are released by the University)
• Get podcasts (on iPhone you can do that directly via iTunes, on other handsets stream to browser)
• Postbox (inc. collection times)
• Find nearest library holding a specific item (queries library catalogue using Z39.50)

To get Map locations in a mobile browser is a challenge – at the moment you need to have one of the following:

• Support for HTML5
• Support for GoogleGears
• Blackberry (can enable access to GPS in settings)