Automatic reasoning evaluation in diet management based on an Italian cookbook

In this paper we consider the problem of the quantitative and qualitative evaluation of an automatic reasoner for the diet domain. In [3, 17] we presented an initial quantitative evaluation of a dietary assistant based on simple temporal problems (STPs). We evaluated our system with a simulation based on hospital menus. However, often daily life menus are not controlled as in hospital, where patients can eat only food approved by doctors. In order to produce a wider realistic evaluation of the STP reasoner, we build a new set of menus based on the Gedeone recipe book, that is a collection of traditional Mediterranean food, and we use these menus in a new simulation experiment. We also introduce a smartphone app implementing the dietary assistant and we report a first human-based qualitative evaluation of such a dietary assistant provided by a professional dietitian.