Find out more about Carbomap here: About.
This week the EEO/AGI-S seminar welcomed Lancaster University Professor of Digital Humanities and Edinburgh GIS alumnus Professor Ian Gregory. Professor Gregory’s work sits at the fascinating and relatively unexplored intersection of GIS, humanities and computational linguistics.
Firstly Professor Gregory explained his work using GIS in literary analysis. The hope is that by using GIS we can reveal changing perspectives over time and gain a greater insight into historical text. Using the writings of Thomas Grey and Samuel Taylor Coleridge in the Lake District, his team georeferenced the texts (linking the place names mentioned to the geography of the Lake District) which enabled the subsequent exploration and study of the author’s individual perspectives. Specifically they could map and understand where the authors were interested in and their mood (what descriptions they tended to use for different landscapes and features).
Secondly Professor Gregory talked of how text can increase the explanatory power of spatial analysis. Using childhood mortality data from the 19th century he illustrated some of the weaknesses of spatial analysis. Although the data can tell us where and how many children died in a given region in a given year, it offers few clues as to why they died, or explanations for the different rates of change in different areas. The team used newspapers contemporary with the data to try to explain the patterns. Using collocation; the technique of determining what words occur most frequently around a mention of a location, they were able to map mentions of words like cholera and dysentery to see where they were being mentioned in relation to. Consequently these collocations could be mapped and perhaps the patterns better understood with easy reference to the texts when needed. These source text were digitised so vast amounts could be feasibly studied and mapped.
To summarize the overall message; the value in textual GIS is the ability to easily make the jump from quantitative to qualitative explanations of patterns and processes. Geography and textual analysis can provide explanations for historical phenomena where traditional statistical analysis is weak.
(MSc in GIS at University of Edinburgh)
Dr Alex Singleton from the University of Liverpool joined us on the 31st January 2014 to speak on the topic of ‘Advances in Geographic Data Science: Open Data and Systems’. Open geographical information systems were discussed in the context of school pupil travel data and census data by this self-confessed ‘Geodemographics nerd’.
The presentation began by discussing the differences between open data and free data; free data allows the use of interfaces and applications but the data is owned by whoever is running the application (eg. google). Open data gives full access to do whatever they like with the data within the licensing agreements. Dr Singleton also discusses how linking private data but sanitising it for publication in the public domain (such as following a person’s journey through school and higher education) can display patterns that may not be seen otherwise. This is an application of ‘big data’.
Dr Singleton expressed his very strong feelings about potholes in his home city of Liverpool. He showed how the ‘Street Bump’ app may be used to let the council know that a pothole exists. However this would seem to highlight potholes in affluent areas where more people have access to smartphones, thus creating a false big-data set and showing the care needed when dealing with open data.
Dr Singleton then went on to discuss two projects where he has used open data to create new geographical systems. The first investigated emissions generated by pupils’ journeys to and from school. The model was created using ‘R’ and postgreSQL, and incorporated data on the pupils’ home addresses, school addresses, probable routes to school, average emissions of cars in the area and likelihood of what type of transport the pupil would be taking.
The second project that was presented was the 2011 Census Atlas. With census data now being open and much more accessible, the number of different and new ways of visualising the data is has increased. Dr Singleton took all the census data for England and Wales and created a model that automatically rendered the appropriate variables into choropleth maps. Again the model used the ‘R’ programming language to create these from the census database, in conjunction with a latex file to produce a formatted output. The time to render all 134,567 maps was only two days. In conjunction with his feelings on open data, Dr Singleton has posted his code for this on Bitbucket for use by others for any data; Irish census data has already been mapped through his method.
In conclusion, Dr Singleton put forward his hopes that the trend towards more open data will continue, thus allowing for more transparency and peer review of the datasets to occur more easily, pushing this area of science forward. He also stated his belief that teaching scripting and coding to students is important for the future, improving the visualisations of these ever-increasing and complex data sets and in turn increasing our knowledge of the world we live in.
(MSc in GIS at University of Edinburgh)
Edinburgh has been successful in gaining several prestigious Scottish Funding Council Highly Skilled Workforce scholarships for its MSc in Geographical Information Science. The world-leading GIS programme at Edinburgh has been awarded three scholarships for full-time study and one for part-time study, available to Scottish residents and EU citizens. Bruce Gittings, the GIS Programmes Director, commented “We are very pleased to receive these SFC scholarships, which represent a very welcome response by the Scottish Government to the lack of trained professionals in the field of GIS”
Despite industry reports which point clearly to a growing demand for a skilled professional GIS workforce, increasing fees across the higher education sector and a declining number of institutions with the capability to offer resource-intensive GIS programmes has left a reducing number of graduates in the UK.
According to the US Bureau of Labor Statistics, GIS occupations will see growth rates ranging from 7%-54% between 2010-2020. In the UK the industry is estimated to currently employ between 8,000 and 10,000 staff providing GIS products and services, with a further 30,000 to 40,000 using the technology [ UK Location Market Study 2012, ConsultingWhere Ltd. ]. GIS salaries have climbed 20% in the year to Feb 2014 (www.itjobswatch.co.uk), with an average salary of £48,000.
The GIS sector is growing beyond the traditional GIS vendors (most of whom are US-based), local and central government, and the environment, to encompass alternative energy, large engineering companies, archaeology and numerous small GIS startups. Yes the output of UK universities numbers not more than 80-100 GIS graduates per year. If the UK is to remain a leader in GI, the government in London has to follow the Scottish Government in promoting this growing part of the IT sector. A strong university education is the route to effective professionals, with a breadth of skills and experiences which can then be further developed in the work environment. If government does not support students and the universities price themselves out of the market, then the industry will either wither or be forced to develop its own training programmes, which are likely to be rather narrower than existing offerings. There are parallels: accounting firms are now taking school-leavers directly into professional training, avoiding the university sector. While this may have its place for technician-level training in the GI sector, the long-term needs of the industry and individual professional staff would not be best-served.
January saw the kick-off in Italy of a large EU-funded project called i-Locate or indoor/outdoor location and asset management through open geodata. Running for three years and involving 24 partners in ten countries, this project aims to use geographical information technology to locate objects in indoor and outdoor spaces, enabling a number of new location-based services (LBS) with potentially significant economic significance. Case studies involve guiding people and tracking equipment in hospitals, health centres, museums and galleries at various test sites around the European Union. The outcome will be an open-source tool-kit which can be freely deployed.
Bruce Gittings of the University of Edinburgh will work alongside noted spatial scientists John Herring (Oracle), Patrick Hogan (NASA Ames Research Centre) and Bart De Lathauwer (Open Geospatial Consortium) to ensure the project is delivered. Bruce spoke at a well-attended Think-Shop event on the 22nd January as part of the three-day meeting in Trentino in Northern Italy which launched the project.
Members of Edinburgh’s GIS community gathered on 17th January in the Old Library at the School of GeoSciences in Drummond Street, where Professor Jason Dykes, of the giCentre, City University London, delivered a seminar exploring “Cartographic Information Visualisation”. Full of elegant and dynamic design solutions to spatial data problems, the well-attended presentation provided an interesting insight into the work that Professor Dykes and his colleagues at the giCentre produce.
Although design in relation to cartography and data visualisation has been a topic of discussion for some time, developments in technology have allowed designers to become more experimental and has opened up fantastic new ways to display and explore spatial data and cartographic design. The first example of this that Professor Dykes mentioned was in relation to the traditional map legend. Since the days of paper maps, the legend has appeared as a static box explaining the symbology used in the map in an order that the cartographer, or some standard has deemed appropriate. But what is the optimal way to represent the legend, and in how should it be ordered? Arguably this is dependent on the map’s use, which may vary from user to user. Prof. Dykes demonstrated how legends could become dynamic in a digital mapping environment as they can be quickly changed from a hierarchically ordered legend at the side of a map, to a spatial legend where symbology is highlighted at the geographical space in which an example of it occurs.
Professor Dykes went on to discuss how, although the spatial dimension of data is important, some distortion of space is often required to optimally visualise the data. Consider the cartogram and the choropleth map: although the cartogram will distort the geographical location, it is likely that this will convey trends and patterns in the data much better than the choropleth map which is constrained by geography. Treemaps use this idea to display hierarchical data, in a way that can be easily compared, using partial geographies. Professor Dykes demonstrated how these types of hierarchical visualisation can be used to detect complicated spatial patterns at various scales in equally varied applications. One of the most interesting examples of this was an investigation he and colleagues had carried out regarding the potential for ‘name bias in alphabetically ordered ballot papers’. In this example Professor Dykes showed how there can be trends in a dataset that are not apparent when viewed at a coarse scale, or when ordered in a particular way, that become so when viewed at a finer scale, or in a different order. The visualisation also showed how although overall there may not have been any indication that the order of names affected elections in London as a whole, in some peripheral boroughs it is likely that this is the case.
The seminar covered several other visualisation issues that were overcome by clever design, such as Irish migration, a video about which can be viewed here: http://gicentre.org/videos/ireland_od_maps.mov.
In his presentation, Pofessor Dykes alluded to the work of Swiss artist and comedian Ursus Wehrli, who ‘tidies’ art by bringing systematic order to well known pieces. Like these ‘tidied’ works of art, good design for data visualisation brings together art and structure creating something that is both aesthetically pleasing, meaningful and easily understood. Professor Dykes and his colleagues have produced fantastic examples of great cartographic information visualisation, which can be viewed on the giCentre website (http://www.gicentre.net/).
Links to all the resources Professor Dykes referred to throughout the seminar can be found here:https://sites.google.com/site/dykesalba/
(MSc in GIS at University of Edinburgh)