An Artificial Neural Network technique for on-line hotel booking

In this paper the use of Articial Neural Networks (ANNs) in on-line booking for hotel industry is investigated. The paper details the description, the modeling and the resolution technique of on-line booking. The latter problem is modeled using the paradigms of machine learning, in place of standard `If-Then-Else' chains of conditional rules. In particular, a supervised three layers MLP neural network is adopted, which is trained using information from previous customers' reservations. Performance of our ANN is analyzed: it behaves in a quite satisfactory way in managing the (simulated) booking service in a hotel. The customer requires single or double rooms, while the system gives as a reply the conrmation of the required services, if available. Moreover, we highlight that using our approach the system proposes alternative accommodations (from two days in advance to two days later with respect to the requested day), in the case rooms or services are not at disposal. Numerical results are given, where the eectiveness of the proposed approach is critically analyzed. Finally, we outline guidelines for future research.

In this paper the use of Artificial Neural Networks (ANNs) in on-line booking for hotel industry is investigated. The paper details the description, the modeling and the resolution technique of on-line booking. The latter problem is modeled using the paradigms of machine learning, in place of standard `If-Then-Else' chains of conditional rules. In particular, a supervised three layers MLP neural network is adopted, which is trained using information from previous customers' reservations. Performance of our ANN is analyzed: it behaves in a quite satisfactory way in managing the (simulated) booking service in a hotel. The customer requires single or double rooms, while the system gives as a reply the confirmation of the required services, if available. Moreover, we highlight that using our approach the system proposes alternative accommodations (from two days in advance to two days later with respect to the requested day), in the case rooms or services are not at disposal. Numerical results are given, where the effectiveness of the proposed approach is critically analyzed. Finally, we outline guidelines for future research.