CAJ | 학술논문

文章利用马尔科夫模型和协同过滤的思想，解决了对“中国科技论文在线”用户进行实时上下文信息的动态推荐问题。先按时间排序对每一个用户的访问URL抽取，提取出状态转移矩阵，再根据协同过滤中的邻居相似度思想用余弦因子法找出最近邻的N个邻居；当给出某用户的当前访问URL时，推荐给他自身和N个最近邻居可能访问的下一个URL的集合。
문장이용마이과부모형화협동과려적사상，해결료대“중국과기논문재선”용호진행실시상하문신식적동태추천문제。선안시간배서대매일개용호적방문URL추취，제취출상태전이구진，재근거협동과려중적린거상사도사상용여현인자법조출최근린적N개린거；당급출모용호적당전방문URL시，추천급타자신화N개최근린거가능방문적하일개URL적집합。
In this paper, it uses Markov model and collaborative ifltering ideas to solve the problem of dynamic recommendation for the"Scientiifcpaper Online"user context information in real time. First, we can access to each user to extract URL according to the time ordering, and extract the state transition matrix. Second, we use cosine factor identify the nearest neighbors of N based on idea of the similarity neighbors in collaborative ifltering. When given access to a user's current URL, and recommended to its and the N nearest neighbors may visit next set of URL.