摘要
传统的检索系统由于其通用的性质,难以满足不同用户所拥有的不同需求偏好。在社会化标注系统中,信息用户能够主动参与网络信息资源的组织与获取,因而其个性化的需求偏好能够较充分地得以体现。文中基于专家制定的叙词表来确定用户需求可能的语义空间维度,然后对用户标注过程中所运用的标签词汇进行量化,在此基础上,运用较成熟的BP神经网络模型,计算出标签词汇与叙词表之间的关联权重,进而用该权重矩阵来表示用户的语义性偏好特征,该语义模型能够在语义空间中定量化地确定信息用户的偏好向量。
It is difficult to meet the different demand preferences for different users in the traditional retrieval system because of its universal nature. However, information users can initiatively take participate in the organization and access to networks information resources in the social tagging system, and therefore the personal preferences can be more fully reflected. Based on the thesaurus developed by experts to determine the dimension of semantic space about user preferences, and then quantifying the tags are by user, the associated weight between tags and thesaurus is calculated by the more mature theories of BP neural network model, and then the weight matrix to express the user semantic preferences is used. The semantic model can quantitatively determine the user preference vector in the semantic space.
出处
《南京航空航天大学学报》
EI
CAS
CSCD
北大核心
2009年第1期139-144,共6页
Journal of Nanjing University of Aeronautics & Astronautics
关键词
用户偏好
语义空间
神经网络
叙词表
user preference
semantic space
neural network
thesaurus