This KTP project was particularly interesting for Sebastian, myself and for the KTP organisation as a whole and we believe it will emerge as one of the most important KTP funded projects - both for cloudBuy and their clients - in recent years.
The project was particularly exciting as it allowed scope for considerable joint work with leading academics at the University of Reading and Research Scientists at cloudBuy. Initially it was expected that academics at Goldsmiths - with experience in [stochastic, multi-agent] search and formal methods/program slicing - would inform the project with respect to the use of AI in intelligent web-spidering; however part-way through the project it's direction evolved to focus more on 'spend analysis' :-an important commercial problem at the heart of cloudBuy's business which involves data normalisation.
The system that finally emerged form the KTP project - Spend Insight - was based on a novel database of products and concomitant AI matching engine. Information from client purchase order systems is entered into the system and intelligently matched against the product database. At the heart of this is a system for automatic product attribution based on classification; in essence, automatic ontology generation for each product class, thus allowing variations in pricing and contracting opportunities to be identified.
The KTP academics working on this project at Goldsmiths believe the work to be both (a) extremely academically novel and (b) highly commercially advantageous to cloudBuy and its clients and anticipate it will have a significant effect on future cloudBuy revenue streams and remain very proud of their association with the work.