【摘要】 针对协同过滤推荐中数据稀疏性与网络结构表达能力不足的问题，该文将形式概念分析与复杂网络分析相结合，提出一种基于相似度空间和模糊截集的协同过滤推荐算法。首先，通过定义加权融合相似度，构建对象网络空间与属性网络空间；其次，将相似度关系映射为模糊关系，在对象网络空间中结合节点重要性与对象相似度模糊截集，构建以专家节点为核心的模糊相似社区；最后，在专家社区对应的属性网络空间中利用属性相似度模糊截集进行推荐。在7个数据集上与11种推荐算法的对比实验表明：所提算法在精确度、召回率和F1值3项指标上均显著优于对比算法，它们分别平均提高了21.50%、26.66%和20.52%,且该文算法在5个数据集上取得最优值。更多还原

【Abstract】 To address the issues of data sparsity and insufficient network structure representation in collaborative filtering recommendations, this algorithm combines formal concept analysis with complex network analysis to propose a collaborative filtering recommendation algorithm based on similarity space and fuzzy cut set. Firstly, a weighted fusion similarity is defined to build the object and attribute network spaces. Secondly, the similarity relationship is mapped into fuzzy relations, where node importance and object similarity fuzzy cut set are combined in the object network space to establish expert node-centered fuzzy similarity communities. Finally, the attribute similarity fuzzy cut set is applied in the attribute network space corresponding to the expert community to make recommendations. Comparative experiments on seven datasets with eleven recommendation algorithms show that the proposed algorithm significantly outperforms the baselines in terms of precision, recall, and F1-score, with average improvements of 21.50%,26.66%,and 20.52%,respectively. It also achieves the best performance on five of the datasets.更多还原