The Methodius Natural Language Generation systems generates personalized descriptions of objects from a collection. As part of the user modeling component, it creates comparisons between the current object being viewed and previous objects from the user history. We present our general algorithm for choosing the best comparison, which can be optimized to give the best result for different domains through a parameterizable scoring function.
|Title of host publication||Proceedings of the International Workshop on Personalization Enhanced Access to Cultural Heritage (PATCH07)|
|Number of pages||11|
|Publication status||Published - 2007|