Building adaptive systems: a neural net approach.

TL;DRAbstract

This chapter introduces a methodology that assists in making tourism counselling and recommender systems 'adaptive'. The kind of 'adaptivity' desired here is based on a system's capability of learning about changes in three respects: (i) macro patterns of consumer information acquisition and usage; (ii) correspondence between the (rather fuzzy) user language and the (more technical) jargon of product descriptions; and (iii) the way of exploiting the users' choice decisions to make inferences about the decision relevance ('weights') of tourism and leisure product attributes. The DieToRecs system prototype serves as a practical and empirical example. The DieToRecs project tried to recognize different macro patterns (named user 'decision styles') that are likely to require particular versions of system functionality. Neural network techniques were used to provide the various learning capacities and to complement the case-based reasoning approach adopted for the DieToRecs project.

Chat with Paper

AI Agents for this Paper