The 'path model' is an approach to representing and accessing information. It is complementary to traditional information retrieval techniques -- and perhaps even complementary to pretty much all of computer science. (No, really.) It extends collaborative filtering, using patterns in logs of user activity. Systems based on paths dynamically classify and structure information on the basis of the ongoing activity of each of a community of users. I've developed systems for heterogeneous data such as web pages and program files, and I'm looking now to expand to new application areas, for example music files/CDs, and artefacts in museums and galleries. In April I'll do a Cakes Talk on a system based on the path model supporting information access and software reuse. In this talk, though, I'd like to focus on the origins of the path model and its more hardcore theoretical aspects.
The path model was partly based on an analogy with a theory of urban structure and development, Hillier's 'Space Syntax'. Another big influence was a number of books on cartography, semiology and philosophy. One aspect of this work is a view that perception of information representations is affected as much by 'high-level' perception -- language and semiology -- as by 'low-level' issues such as visual resolution and colour discrimination. I'll give examples of how architectural and information structures are to some extent equivalent, and how the path model conforms to contemporary poststructuralist semiology in the way that systems represent information and use -- in particular adaptive, subjective representations encompassing heterogeneous data. You can find papers on the system design and theoretical sides of this work on my web page: http://www.dcs.gla.ac.uk/~matthew.
For more information contact: matthew@dcs.gla.ac.uk