The Natural Browsing project and PASCAL - a rather unique collaboration

On the face of it, involvement with the dissemination of a leading-edge computer science project would seem to be outside of PASCAL's remit.  What have "semantic search" and "natural languague processing" got to do with the social sciences? This is exactly the question we asked ourselves when invited by our colleagues from the Puglia Region to get involved in the Natural Browsing project. What we discovered was that these concepts and the technology being developed behind them hold a great deal of promise for organisations like PASCAL in assisting us in achieving our primary functions.

The Natural Browsing project aims to develop "a framework able to automatically build a knowledge base from unstructured  data". The project relies on natural languague processing methods and semantic web technologies in order to achieve this. In a nutshell, much of the data on the world wide web ("WWW") relies on simple clickable links from one chuck of information to another. Recently, search facilities (such as those provided by Google) present us with a number of related links based on patterns of similar searches and to some extent our personal search habits. The Natural Browsing project is working towards taking these related links a step further - to attempt to organise content based on meaning rather than just facts and similarly match our enquiries by an interpretation of the meaning of our requests.

PASCAL, in it's role as an international observatory collects information about local, regional and national developments in place management, lifelong learning and social capital, to provide informed opinon and conclusions which help guide policymakers on a worldwide scale.  Often the relationships we explore depend more on the knowledge and experience of our Associates and Affiliates more than the prima facie situation might indicate.  Many of us use the internet to support this process, so the concept of "assisted" search facilities which attempt to "understand" what we are looking for is nothing short of the equivalent of having a team of (very well-informed) student PhDs on the team!

What we produced as our contribution to the dissemination of the Natural Browsing project has been developed at PASCAL's European Centre at the University of Glasgow with assistance from Steve Rubin, responsible for the development and technical support for a wide range of PASCAL internet facilities.  Our goal was to provide a non-technical explanation of the concepts embodied in the Natural Browsing project through a knowledgebase of our own and help educate colleagues on the principles involved in deriving meaning from data.  We all perform the task of deriving meaning every waking minute of the day, but describing exactly how we do that is a different proposition altogether. It is our hope that a better understanding of the machine processes involved in mimicking this natural human task will help us all better formulate the questions we ask and even improve the way in which we record our findings.

As our project developed it became obvious that we were creating a very crude approximation of a semantic knowledgebase which could be used by the Natural Browsing development team to educate others, test sample processes and provide comparisons with existing technologies. So in a way, the shortcomings of our contribution from the perspective of embodying meaning is also a good example of the need for the work of the Natural browsing project!

In summary, I recommend even the most technology-adverse colleagues to make the effort to explore the PASCAL-sponsored Natural Browsing wiki, if for no other reason than obtaining a non-technical explanation of some of the most important concepts behind the way in which information is being organised for presentation to us Luddites.

Mike Osborne