摘要
知识发现领域中,个性化推荐技术因其应用广泛受到了业界的广泛关注和高度重视。但由于用户隐私保护方面的限制,现有的推荐系统不能直接挖掘用户的个人信息,因此只能采用表征用户爱好的特征值来间接地挖掘用户信息。针对此类问题,提出了一种新的推荐方法。该方法可自动提取相应领域的特征值,并基于领域关键词过滤冗余的领域特征值,从而据此构建用户偏好模型,并与协同过滤算法绑定,生成最终的推荐结果。为验证所提出推荐方法的有效性和可行性,基于实时数据集与其他已有的推荐方法进行了对比实验,并基于对比实验结果进行了相关的分析研究。对比验证实验结果及其分析表明,该推荐方法能够有效地提取领域特征值,其推荐的精准度也有所提高。
In knowledge discovery, personalized recommendation technology has received extensive concern and high attention because of its wide application. However, due to the limitations of user privacy protection, the existing recommendation system can＇ t directly mine the user＇ s personal information. So,the features which imply user preference to indirectly mine user information can be utilized. In order to solve above problem, a new recommendation method is proposed which can automatically extract relevant domain features and filter the redundant domain features based on domain keywords to construct a user preference model and generate the final recommendation result in combination with the collaborative filtering algorithm. To verify its effectiveness and feasibility, compared with other existing recom- mendation methods based on a real time data sets the experiments for verification are conducted. The results of contrast experiments and relevant analysis show that it can effectively extract the domain features and its accuracy of the recommendation is improved.
出处
《计算机技术与发展》
2017年第11期88-91,共4页
Computer Technology and Development
基金
国家自然科学基金资助项目(61100213)
南京邮电大学教育部重点实验室开放研究基金(ZS035NY11005)
关键词
领域特征值
协同过滤
用户偏好模型
个性化推荐
domain features
collaborative filtering
user preference model
personalized recommendation
