【摘要】 为使用户在手持设备上提取用户兴趣是重要基础,提出一种基于用户隐式反馈的方法。该方法从网页中提取兴趣块(用户感兴趣的网页内容),假设用户的浏览行为如在块上的停留时间、滚动次数、滚动速度、进入链接的次数等与用户是否对块感兴趣相关。从手持设备上的浏览行为中提取了34种隐式反馈特征,通过分析18个用户在608个网页的9474个块上的浏览行为,验证了其中29种特征与兴趣块的相关性,并且这种相关性能用于跨用户、跨网站的兴趣块提取。该文的研究成果可用于建立个性化的用户偏好模型,应用在手持设备上的自适应网页浏览中。更多还原

【Abstract】 Personalized web page adaptation on handheld devices is based on learning the user’s interests.A method was developed to identify a user’s interest blocks using implicit feedbacks extracted from the user’s browsing behavior,such as display time,scrolling and link selection.34 kinds of block-based implicit feedbacks were extracted,29 of which prove to be statistically relevant to interest blocks through a user study with 18 participants who browsed 9 474 blocks on 608 pages.C4.4,SVM,and NaiveBayes methods were used to build a model from a combination of 29 relevant feedback indicators to identify interest blocks.Experimental results show the effectiveness of the learning model and that the precision and recall are affected by each user’s habits and the web sites.The identified interest blocks can be used to identify user preferences for personalized adaptive web browsing on handheld devices.更多还原