论文部分内容阅读
为解决当前个性化推荐不能同时考虑用户兴趣点受时间和频率两个维度共同影响问题,提出一种基于本体相似度和时间衰减的动态个性化推荐算法。首先确定时间衰减函数来计算用户兴趣点的时间衰减规律,然后考虑不同兴趣点访问频率对兴趣点关注程度的影响,再次通过本体相似的综合算法解算推荐资源与关注目标资源的相似度,最后通过兴趣点更新权重与相似度的加成效果确定推荐资源的排序。实验结果表明该算法在整体上具有较好的综合性能,推荐资源的查准率和查全率均较高,算法的普适性和效果较好。
In order to solve the problem that the current personalized recommendation can not simultaneously consider that the user’s interest point is affected by the two dimensions of time and frequency, a dynamic personalized recommendation algorithm based on ontology similarity and time attenuation is proposed. Firstly, the time decay function is determined to calculate the time decay rule of the user’s interest point. Then, considering the influence of the frequency of different points of interest on the degree of interest of the point of interest, the similarity of the recommended resource and the target resource is solved by the ontology-like comprehensive algorithm again. The ranking of recommended resources is determined by the addition of interest points updating weight and similarity. Experimental results show that the proposed algorithm has good overall performance, and the accuracy and recall of recommended resources are high, and the algorithm has better universality and effectiveness.