摘要
传统医院图书馆数据库文本分类是对指定规模文本的分类,无法针对特定用户的浏览内容实施分类。为此,提出基于协同过滤的医院图书馆数据库文本分类优化方法。将用户浏览数据库文本的特征看做物品,构建半自动编码器的协同过滤模型优化用户物品评分矩阵,使用平均评分修正因子、热门物品惩罚因子改进相似度计算。引入注意力机制构建CNN-SVM分类模型,将用户文本浏览特征作为输入,实现文本分类。测试表明,该方法构建评分矩阵的RMSE最低,推荐图书馆文本阅读特征的MAE值最小,在文本分类上F1值达到96.5%,有较好的文本分类效果。
Traditional hospital library database text classification depends on specified size,which cannot be implemented based on the browsing content of specific users.Therefore,a text classification optimization method for hospital library database based on collaborative filtering is proposed.The characteristics of users browsing the database text are regarded as items,and a semi-automatic coder collaborative filtering model is constructed to optimize the user item scoring matrix.The average score correction factor and popular item punishment factor are used to improve the similarity calculation.Introducing attention mechanism to construct a CNN-SVM classification model,which takes user text browsing features as input to achieve text classification.Tests have shown that this method has the lowest RMSE for constructing a rating matrix,the lowest MAE value for recommended library text reading features,and an F1 value of 96.5%in text classification,demonstrating good text classification performance.
作者
从莉萍
沈剑文
王海生
CONG Liping;SHEN Jianwen;WANG Haisheng(Affiliated Mental Health Center of Shanghai Jiao Tong University School of Medicine,Shanghai 200030,China;Eye and ENT Hospital of Fudan University,Shanghai 200030,China)
出处
《微型电脑应用》
2024年第2期146-148,153,共4页
Microcomputer Applications
关键词
协同过滤
医院图书馆
数据库
半自动编码器
文本分类
collaborative filtering
hospital library
database
semi-automatic encoder
text classification
