改进的贝叶斯算法在旅游景点推荐中的应用.doc

改进的贝叶斯算法在旅游景点推荐中的应用 摘 要：信息技术的飞速发展使得旅游信息呈爆炸式增长根据游客的特征和偏好已十分必要。MI-NB算法是在传统贝叶斯分类算法不足的基础上进行改进的，应用了互信息的知识，通过相对可信度R来进行特征选择，以删除冗余属性，并把R作为权值引入到NB算法中，从而得到改进后的MI-NB算法。采用改进的贝叶斯分类算法(MI-NB)来能够大大降低分类数据的维数，有效提高了景点推荐的准确率。 关键词：特征选择；互信息可信度分类景点推荐 中图分类号：TP393 文献标识码：A Improved Bayesian algorithm recommended in the intelligence recommendation System of Traveling InformationLINGHU Hong-ying1, JIANG Ji-chun2 (1 Vocational and Technical College, Guizhou Normal Colleg e, Guiyang, Gu izhou, 550018 Ch ina; 2 College of Computer Science and Information, Guizhou University, Guiyang Guizhou,550025,China) Abstract: With the rapid development of information technology, tourism information has become more and more huge. Thus, according to the visitors’ characteristics and preferences, the research and design of an intelligence recommendation system of traveling information has been imperative. MI-NB algorithm based on the shortage of traditional Bayesian classification algorithm and the mutual information knowledge is proposed in this paper. The relative credibility R is considered as the standard of feature selection to remove redundant attributes and is introduced into NB algorithm as the weight to obtain the improved MI-NB algorithm. Experiments showed that MI-NB algorithms could greatly reduce the dimension of data and effectively improve the accuracy of recommendation of tourist attractions. Key words: feature selection; mutual information; credibility; Bayesian classifier; attractions recommendation 0 引 言 随着国内经济的快速增长，我国旅游行业也迅速发展，并在我国这样的内需型经济生活中发挥着重要的作用，成为国民经济中新的经济增长点，国内经济的发展促进旅游业繁荣，随着旅游休闲的观念逐渐深入人心，旅游消费也势必成为消费升级的必然之选，旅游业的发展在国民经济中所起的作用也越来越重要。然而就旅游行业目前的发展趋势而言，游客和旅游机构的供需之间存在着一个“沟通真空”。这个“沟通代沟”主要体现在：一方面旅游者的消费特征更加个性化，对旅游机构提供的服务也越来越挑剔，面对Internet上每天大量全新的旅游资源，游客需要旅游机构推荐满足其兴趣爱好、实际需求的旅游信息。另一方面，大多旅游机构依旧停留在传统的旅游业经营模式上，服务项目单一，旅游线路、旅游景点的介绍几乎都是固定的模式，多以文字和图片等传统方式进行粗浅地描述。 为了解决上述问题，本文采用基于协同过滤的推荐算法来进行旅游景点的推荐。基于协同的推荐算法要求充分考虑游客之间的相似性，把游客进行分类，为同一类的游客推荐同一类景点，所以，基于协同过滤推荐算法可以用分类算法来实现。旅游数据具有海量性、高维性、不完全性、多