To further increase the successful recommendation rate of a ubiquitous hotel recommendation system, an incremental learning and integer-nonlinear programming approach (INLP) is proposed in this study to mine users’ unknown preferences. In the proposed methodology, an INLP problem is solved to adjust the values of weights in the recommendation mechanism to be closer to those in the decision-making mechanism so as to maximize the successful recommendation rate. In addition, the weights are adjusted on a rolling basis so that more historical data can be considered without inflating the INLP model. The experimental results showed that the proposed methodology outperformed several existing methods in increasing the successful recommendation rate, even with a cold start.
|Number of pages||10|
|Journal||Journal of Ambient Intelligence and Humanized Computing|
|State||Published - 1 Jul 2019|
- Incremental learning
- Integer-nonlinear programming
- Ubiquitous recommendation