基于模糊角分类的神经网络用户兴趣模型分类算法

用户兴趣描述文件的快速分类是个性化搜索引擎的关键技术，提出了一种模糊角分类神经网络模型，该模型能接受用户兴趣描述文件的实向量输入，克服了角分类神经网络（CC4）对二进制输入的要求。模糊角分类神经网络模型根据用户信息所落入的k最近邻的样本泛化空间来进行分类，随着k值的增大，其分类效果趋近于贝叶斯分类算法。

User interest profile classification algorithm based on FCC neural network

Fast classification of user interest profile is a key technology for personalization search engine. A new kind of FCC neural network model was presented in this paper. It does not need the binary system input that is usually required by CC4, because it can accept the real vectors input. FCC neural network model works according to the k-nearest neighbor samples' generalization space which users' information falls into. With the increasing of value k, the classification effect becomes close to that of Bayes classification algorithm.