利用最近邻域推荐且结合情境感知的个性化推荐算法

Personalized Recommendation Algorithm Based on Recent Neighborhood Recommendation and Combined with Context Awareness

针对传统情境感知推荐算法推荐精确度低和适用环境受限等问题,提出了一种可行的解决方案。该方案可以根据检测到的情境信息找到相关的媒体内容,比仅依赖特征提取的方案更有效。首先,利用情境数据和搜索信息来识别所选项的情境与特定情境中用户的兴趣度之间的隐藏关系,并构建未知排名的推荐模型。然后,通过使用给定的情境列表来计算用户对项目的预期排名分数,从而进行情境感知评级。根据用户的情境参与选择新项目,从而使检测到的情境有助于促进对相关项目的搜索。进一步使用优化函数来最大化结果推荐的平均精度(MAP)。实验结果表明,与目前较为先进的两种算法相比,提出的方法表现出了比传统协同过滤算法更好的性能,且分别使平均绝对误差值降低了1.8%和1.2%,在推荐精确度和召回率方面也均优于两种对比方法。

Aiming at the problem that the traditional context-aware recommendation algorithm is not accurate and the applicable environment is limited,this paper proposed a feasible solution.It can find relevant media content based on the detected context information,which is more effective than relying solely on feature extraction.First,the context data and the search information are used to identify a hidden relationship between the selected context and the user’s interest in a particular context,and a recommendation model of the unknown ranking is constructed.The contextually perceived ra- ting of the user’s expected ranking score is then calculated by using the given contextual list.The user’s situation is used to participate in the selection of new items so that the detected situation helps to facilitate the search for related items .The optimization function is further used to maximize the average accuracy (MAP) of the result recommendation.The experimental results show that compared with the more advanced algorithms,the proposed method shows better performance than the traditional collaborative filtering algorithm,and the absolute error value is reduced by 1.8% and 1.2% respectively in the recommendation accuracy and recall.The rate is also superior to the two comparison methods.