Basic Consideration of Collaborative Filtering Based on Rough Co-clustering Induced by Multinomial Mixture Models
Basic Consideration of Collaborative Filtering Based on Rough Co-clustering Induced by Multinomial Mixture Models
复制标题
DOI：
10.1109/scisisis55246.2022.10001887
复制
发表时间：
2022-11
期刊：
影响因子：
--
通讯作者：
S. Ubukata;Kenryu Mouri;Katsuhiro Honda
中科院分区：
文献类型：
--
作者：
S. Ubukata;Kenryu Mouri;Katsuhiro Honda
协同过滤（CF）是一种基于用户的偏好向每个用户推荐优选内容的方法。聚类是一种通过提取由相似对象组成的簇来对数据进行自动分类和汇总的方法。粗糙聚类是基于粗糙集理论，考虑对象属于类的不确定性，提取重叠类。共聚类是一种用于总结对象和项目之间的共现信息的有用技术。由于CF任务中的数据集被认为是包含基于人类感官和敏感性的不确定性的共现信息数据，因此粗糙共聚类被认为适用于CF任务。在这项研究中，我们提出了协同过滤的基础上粗糙的共同聚类诱导的多项式混合模型（RCCMM-CF）。此外，我们通过使用两个真实世界数据集，即NEEDS-SCAN/PANEL数据集和MovieLens 100 K数据集的数值实验，验证了所提出的方法，RCCMM-CF的推荐性能。
Collaborative filtering (CF) is a method for recommending preferable contents to each user based on the user’s preference. Clustering is a method for automatically classifying and summarizing data by extracting clusters composed of similar objects. Rough clustering can extract overlapped clusters by considering the uncertainty of belonging of object to clusters based on rough set theory. Co-clustering is a useful technique for summarizing co-occurrence information between objects and items. Since dataset in CF tasks are considered to be co-occurrence information data that include the uncertainty based on human senses and sensitivities, rough co-clustering is considered to be suitable for CF tasks. In this study, we propose collaborative filtering based on rough co-clustering induced by multinomial mixture models (RCCMM-CF). Furthermore, we verified the recommendation performance of the proposed method, RCCMM-CF, through numerical experiments using two real-world datasets, namely, NEEDS-SCAN/PANEL dataset and MovieLens 100K dataset.