Artificial Intelligence Support for *The User around the Marketplace*: Automatic Engineering of Interactive E-commerce Applications

When thinking about opportunities for e-commerce, it is necessary to differentiate scenarios in which the user's aims and attention are close to the items that may be offered to him (e.g. when he has just entered an online bookshop) from scenarios in which they are not (e.g. in delivering publicity while he watches TV). The support available for the former is quite advanced nowadays, especially thanks to the development of recommender systems. On the contrary, in cases in which the user gets to know about interesting items casually, there is much place to provide him with better guidance and thereby increase the number of transactions accomplished online. The key aspects have to do with contextualizing the selection of the items that will be offered to the users, delivering the information they may find more valuable and understandable in the most accessible way, and identifying the most convenient providers taking into account not only price and availability but also the users' location. We describe a system that furnishes these features, relying on semantic and rule-based reasoning techniques to automatically compose interactive services that let the users browse details of (and purchase online) various kinds of items they might be interested in, as inferred from knowledge about their individual preferences, interests and needs. A prototype targeted at improving Digital TV advertising is described.