摘要
本文首先利用有限混合模型对前向神经网络的交叉覆盖算法进行优化,再将优化后的覆盖算法应用于已进行分词预处理的中文文本数据库。从实验结果来看,优化后的覆盖算法在测试精度上取得了令人满意的结果,10次实验所得到的平均精度除经济类外,其余都不同程度高于原覆盖算法处理同类数据的分类精度。
This paper firstly optimizes the intersecting Cover Algorithm (CA) of forward neural networks with the finite mixture model. Then, the paper applies the optimized CA in the Chinese text corpus which has been pre- processed into cut words. The experimental results show that the optimized CA has achieved satisfied results in test precision. The average precision of the 10 experiments is higher than that of the original CA to certain extent except the precision of the Economic Category in processing the same database.
出处
《情报理论与实践》
CSSCI
北大核心
2010年第5期92-94,128,共4页
Information Studies:Theory & Application
基金
安徽省哲学社会科学规划基金(项目编号:AHSKF0708D13)
安徽省教育厅人文社会科学研究基金(项目编号:2009sk038)的研究成果之一
关键词
信息检索
模型
覆盖算法
information retrieval
model
cover algorithm