Sakai Applications: Social Science

Social Science Applications

The SABRE-R e-Social Science pilot project aims to port the open-source SABRE statistical analysis package to a parallel computing Grid-based environment. Middleware extensions to the functional R programming language are being added using Web services in the GROWL library and appropriate additional wrappers for R clients and services This has already been demonstrated for the simplest cases.

This pilot project will play a key role in the development of some of the middleware components and services appropriate to e-Social Science. Our approach has a number of distinct advantages for the social scientist because it will:

1. promote the re-use of existing software (by wrapping it as components), which can easily be incorporated into other analyses; 2. greatly simplify programming to the level where at which applications may be constructed simply by plugging components together (e.g. using a graphical build tool), or by using an appropriate scripting language; 3. allow sophisticated management of the underlying distributed/ networked environment, which can be separated out from the work of the social scientist.

Such a modularised approach to software development is sufficiently flexible to cope with many different scenarios. It will open up any number of opportunities, ranging from Grid-enabled data fusion and visualization to computation and simulation.

In this project we are also applying the GROWL middleware to wrap the statistical modelling methods for analyzing work/ life history data, and to make these developments available in the distributed environment as a componentised R library. The free-to-use R language and environment provides a wide variety of useful statistical and graphical techniques (linear and non-linear modelling, statistical tests, time series analysis, classification and clustering).

To make R even more user friendly for social scientists, an open source GUI interface, R Comander, has been developed. The menu lists in R commander will be edited to include GROWL and SABRE-R giving a simple point and click interface to remote data and applications on the National Grid Service clusters. We will also evaluate this interface for the bio-informatics applications.

The evaluation community includes a large number of Lancaster University researchers from a variety of social science disciplines that have demonstrated excellence in research by their ratings at the last RAE. These include: Applied Statistics (5*), Management (5*), Sociology (5*) and Computing (5). They are focussed on the Centre of Excellence for e-Social Science and are participating with the National Centre for e-Social Science and collaborating with a wider group of UK researchers through first hub CQeSS, which is a Collaboration for Quantitative e-Social Science.

Activities and Deliverables