论文部分内容阅读
针对科技文献特征词在语义上的层次特性,提出基于概念泛化的内容过滤推荐算法。采用矢量空间模型作为用户兴趣偏好和科技文献特征的描述模型;在比较科技文献特征与用户兴趣偏好的相似程度时,首先从字符层面比较科技文献特征词与用户兴趣特征词,然后在基于ODP目录结构的用户兴趣偏好概念泛化树上对字符不相同的特征词对进行语义比较,并修正特征词权重,以避免遗漏“字符不同,但语义相似”的关键词对。理论分析和实验结果表明,该算法能够更加全面、准确地推荐科技文献对象。
According to the semantic features of feature words in scientific literature, a content filtering recommendation algorithm based on concept generalization is proposed. Using the vector space model as a description model of the user’s interest preferences and the features of the scientific literature; when comparing the similarities between the features of the scientific literature and the preferences of the user, the feature words of the scientific literature and the user interest feature words are firstly compared from the character level, The structure of the user interest preference concept on the generalized tree features different characters on the word pairs of semantic comparison, and the amendment of the weight of feature words in order to avoid missing “different characters, but semantic similarity ” key words. Theoretical analysis and experimental results show that this algorithm can comprehensively and accurately recommend scientific literature objects.