A collaborative filtering recommendation system for e-tourism from a specific cultural perspective

• Main Authors: Daniel Silva Navarro, 辛丹尼
• Other Authors: 許永真
• Format: Others
• Language: en_US
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/87417717601423145330

碩士 === 國立臺灣大學 === 資訊工程學研究所 === 102 === Traveling is a very important activity on human life; moreover, is a very profitable business all over the world. However, when people start planning their trips overseas, searching for information about places to visit, can be a very time consuming and misleading task. This research, aims to create the first module of a bigger e-tourism recommendation platform for Taiwan travelers. This initial module, will focus in Taipei pre-travel issues and it will recommends point of interest using a collaborative filtering approach. To effectuate the recommendation, a modified version of slope one algorithm was utilized to predict the unavailable ratings on the dataset and mixed with the traditional CF prediction algorithm. This mixed algorithm showed a 10.19\% MAE improvement in comparison to the basic traditional collaborative filtering approach. To effectuate the experiments for this recommendation system, the dataset is composed by the most popular 86 points of interest in Taipei. These point of interest were reviewed by 27 foreigners living in Taiwan.