基于协同过滤的酒店推荐系统研究与实现

【作者】 刘艳；

【导师】 张凤荔； 马建立；

【作者基本信息】 电子科技大学 ， 软件工程， 2011， 硕士

【摘要】 Web2.0时代,高速增长的信息量已经远远超出了人们处理它的能力。电子商务网站和用户都面临着信息过载的难题,这就迫切需要人们采用一定的手段从海量的信息中找出最有价值的信息,于是推荐系统应运而生。目前,推荐系统和推荐技术已经成为互联网最热门的研究领域之一。本文就是在这样的背景下,以某旅游电子商务网站为实例,进行酒店推荐系统的研究。本文根据网站的产品形态,围绕酒店搜索结果页和酒店展示详情页设计整个推荐系统,针对两类页面的不同特点,采用不同的推荐方式。对搜索结果页,在分析了用户搜索行为之后提出了基于酒店点击转化率的启发式算法;对酒店详情页,从经典的协同过滤出发,以资源对资源算法入手,研究协同过滤方法在酒店详情页中的应用。最终的目的是提高酒店预定的转化率,增加网站营收。最后,为了说明本文所设计的酒店推荐系统的有效性,文章根据酒店真实数据对酒店推荐系统进行了验证。更多还原

【Abstract】 In the age of Web 2.0, the amount of information is increasing far more quickly than our ability to process it. Both the E-Commerce sites and the customers face the serious problem of information overload. A new technology that can help us sift through the available information to find which is most valuable to us is in urgent need. Thus, recommender systems have become an important and one of the hotest research area in the internet research. We study the hotel recommender systems based-on an E-Commerce travelling site.The recommender system has two parts, the search page part and the detail page part, according to how we use the site. We propose a heuristic algorithm based on hotel conversion rate for the search page after we look deep into the user’s behavior and the hotel click data. Moreover, we try to apply the collaborative filtering algorithms in detail page recommendations. We also compare the difference and quality between the item to item and item-based collaborative filtering algorithms.Experiments are designed to validate these algorithms. The results prove that the heuristic and the item to item collaborative filtering algorithms were both effective in our hotel data sets. Besides, the website would have big progress in enhancing the conversion rate if these algorithms were implemented.更多还原