On matching agent intents with Web affordances

More and more home appliances and manufactured products have the ability to connect to the Internet. As a consequence, these products can expose Web services that automated Web agents can choose to consume or not (e.g. to turn on a light, regulate ambient temperature, configure an alarm, etc). In the case the number of available Web services is high, matching the relevant Web services to user intents must be (partially) automated.
The thesis should introduce an approach for users to express their intents such that a virtual home assistant can easily match them to available Web services. The latter are assumed to be exposed in the form of machine-readable descriptions, also called ‘affordances’ in the context of the Web. Approaches analogous to offer/demand matching techniques on marketplaces may be applied.