Developing More Intelligent Applications with Data Mining

Developing More Intelligent Applications with Data Mining, by Rafal Lukawiecki - Video presentation from Tech-Ed 2007.

Imagine an application that adjusts its behaviour as it learns about the needs and characteristics of its users: for instance, a user interface that knows what the user is most likely to do, or select next. Or, perhaps, a Web store that can predict a shipment delivery failure even before the order has been placed, and warns the user. How about software that knows it is being abused and is on the brink of being hopelessly attacked by a criminal? In these three cases, and many others, the traditional approach would require you to think of all the circumstances in advance, and to encode them as some form of logic. Naturally, that is an impossible task to complete. Data Mining offers us a glimpse of a world to come: your application will self-trace its past behaviour in order to discover these patterns for you. Your code, in turn, will use the ever-evolving data mining model to perform predictions based on which way the application would react. While the world of academia has been thinking of similar approaches, notably in the Artificial Intelligence research, it has only recently become an everyday possibility to harness the power of Data Mining to a similar extent. Come to this session if you would like to see where the future is headed. This session introduces the key concepts of data mining and its algorithms for the benefit of those not yet familiar with the field, but the bulk of the presentation concentrates on the as yet unexplored world of building more intelligent applications.

Read more...