摘要
非线性遗忘函数能够改进传统协同过滤推荐算法没有考虑游客兴趣稳定性的缺点,从而实现精准旅游个性化服务推荐。但是非线性遗忘函数并没有考虑到旅游产品的流行度,流行度越高的旅游产品,游客之间的兴趣相似度便越不准确,而流行度越低的产品,预测游客之间的兴趣相似度就更加准确。鉴于此,为了更进一步提高推荐精准度,在非线性遗忘函数的基础上构建考虑旅游产品流行度的数学模型,削弱流行度高的旅游产品的权值,调整游客间兴趣相似度。实验表明,引入产品流行度后,得到的平均绝对差值变小,推荐精准度也显著增加。
Ahhough nonlinear forgetting functions can be used to overcome the shortcoming of traditional collaborative filtering recommendation algorithms ignoring tourists' interest stability and to realize accurate personalized service recommendations, it fails to take tourism product popularity into consideration. Tourists' interest similarity degree is increasingly inaccurate in highly popular tourism products while it is increasingly accurate in less popular tourism products. In order to improve the recommendation accuracy, a mathematical model involving tourism product popularity is constructed based on nonlinear forgetting functions so as to weaken the weight of highly popular tourism products and adjust tourists' interest similarity degree, The experiment shows that the mean absolute difference value decreases with recommendation accuracy increasing significantly when product popularity is included.
出处
《山东财经大学学报》
2016年第1期92-98,共7页
Journal of Shandong University of Finance and Economics
基金
国家自然科学基金"双重委托代理下旅游服务供应链激励机制设计"(71201053)
湖南省教育厅优秀青年项目"基于双边非对称信息的合作旅游服务质量生产契约研究"(15B070)
湖南工业大学研究生创新基金项目"旅游个性化服务推荐"(cx1507)
关键词
非线性遗忘函数
产品流行度
协同过滤
个性化推荐
nonlinear forgetting function
product popularity
collaborative filtering
personalized recommendation