Estimation of Personal Preferences on Points of Interest

The goal of our study is to recommend a travel plan that a tourist can change his/her state of mind to positive feeling. A personal preference of a point of interest (POI) differs individually. Moreover, the general impressions and the personal knowledge of a POI influence the personal preference of it. Therefore, it is difficult for a tourist to make a travel plan concerning POIs where a tourist has never visited, or he/she has not visited much. In this paper, we propose a method to estimate personal preferences on the unknown POIs by using the general impressions and the personal knowledge concerning known POIs. We estimate personal preferences concerning unknown POIs through shaping parameters about the impression and knowledge of POIs involved in personal preference about POIs. We have designed a hierarchical Bayesian model via Markov chain Monte Carlo technique for parameter estimation. As for 15 subjects, we have designed the model by using the general impressions and the personal knowledge concerning 84 POIs. They evaluated the validity of the results of which the model has estimated individual preferences for unknown POI. As for nine of 15 subjects, the result shows that the estimation accuracy is more than 60%. It was shown that the general impression and the personal knowledge of POIs affected the preference.