[1]王瑞洋,李 涛*.融合药品语义的混合推荐算法[J].计算机技术与发展,2021,31(12):141-147.[doi:10. 3969 / j. issn. 1673-629X. 2021. 12. 024]
　WANG Rui-yang,LI Tao.Hybrid Recommendation Algorithm Based on Drug Semantics[J].,2021,31(12):141-147.[doi:10. 3969 / j. issn. 1673-629X. 2021. 12. 024]
点击复制

融合药品语义的混合推荐算法()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
31
期数:
2021年12期
页码:
141-147
栏目:
应用前沿与综合
出版日期:
2021-12-10

文章信息/Info

Title:
Hybrid Recommendation Algorithm Based on Drug Semantics
文章编号:
1673-629X(2021)12-0141-07
作者:
王瑞洋; 李 涛*
武汉科技大学 计算机科学与技术学院,湖北 武汉 430065
Author(s):
WANG Rui-yang; LI Tao
School of Computer Science and Technology,Wuhan University of Science and Technology,Wuhan 430065,China
关键词:
矩阵分解; 数据稀疏; 药品; 语义知识; 卷积神经网络
Keywords:
matrix factorization; data sparsity; drug; semantic knowledge; convolution neural network
分类号:
TP391
DOI:
10. 3969 / j. issn. 1673-629X. 2021. 12. 024
摘要:
传统的矩阵分解算法对药品进行推荐时,由于存在数据稀疏性问题,导致推荐结果不准确。 因此提出了一种融合药品语义的混合推荐算法( H-DS) 。 首先利用药品的类别信息构建出药品的分类矩阵,从而计算出药品的类别关联度;然后预处理药品主治功能的描述文本,使用卷积神经网络进行训练,得到其对应的特征;最后用概率矩阵分解算法结合药品类别和功能两方面的语义信息来改进模型,修正矩阵分解的项目隐因子特征,从而实现了对药品的精准推荐。 实验表明,在 MAE 和 RMSE 评价指标上,H-DS 较传统的概率矩阵分解算法( PMF) 误差降低了 6% ~ 7% 左右;与卷积矩阵分解(ConvMF) 等 PMF 经典改进模型相比误差也降低了 2% ~ 3% 左右。 与上述算法相比,该算法充分利用了药品的语义信息,可以有效缓解数据稀疏性,在进行药品推荐时效果更好。
Abstract:
When the traditional matrix factorization algorithm recommends drugs,the recommendation results are inaccurate due to the data sparseness. Therefore,a hybrid recommendation algorithm (H-DS) that integrates drug semantics is proposed. First,the category information of the drugs? ? ? ? is used to construct the classification matrix of the drugs,thereby calculating the category relevance of the drugs.Then the description text of? ? ? the drug’s main functions is preprocessed and trained by the convolutional neural network to obtain the corresponding features. Finally, the probabilistic matrix factorization algorithm combines the semantic information of the drug category and the function to improve the model and modify the item hidden factor feature of the matrix factorization,thus realizing the accurate recommendation of the drug. Experiment shows that? ?in terms of MAE and RMSE evaluation indicators,H-DS reduces the error by about 6% to7% compared with traditional probabilistic matrix factorization (PMF) ,and it also reduces the error by about 2% to 3% compared with the classic improved PMF models such as convolutional matrix factorization (ConvMF ) . Compared with the above algorithm, the proposed algorithm makes full use of the semantic information of medicines,which can effectively alleviate data sparsity and has a better effect in drug recommendation.

相似文献/References:

[1]余志虎 戚玉峰.一种基于云模型数据填充的算法[J].计算机技术与发展,2010,(12):34.
　YU Zhi-hu,QI Yu-feng.A Data Filling Algorithm Based on Cloud Model[J].,2010,(12):34.
[2]陈彦萍,王赛. 基于用户-项目的混合协同过滤算法[J].计算机技术与发展,2014,24(12):88.
　CHEN Yan-ping,WANG Sai. A Hybrid Collaborative Filtering Algorithm Based on User-item[J].,2014,24(12):88.
[3]王菲,黄刚,朱峥宇.基于信任聚类的协同过滤推荐算法[J].计算机技术与发展,2019,29(05):22.[doi:10. 3969 / j. issn. 1673-629X. 2019. 05. 005]
　WANG Fei,HUANG Gang,ZHU Zheng-yu.Collaborative Filtering Recommendation Algorithm Based on Trust Clustering[J].,2019,29(12):22.[doi:10. 3969 / j. issn. 1673-629X. 2019. 05. 005]
[4]刘荣权,袁仕芳,赵锦珍,等.基于用户属性聚类与矩阵填充的景点推荐算法[J].计算机技术与发展,2020,30(11):200.[doi:10. 3969 / j. issn. 1673-629X. 2020. 11. 037]
　LIU Rong-quan,YUAN Shi-fang,ZHAO Jin-zhen,et al.Tourist Spot Recommendation Algorithm Based on User Attribute Clustering and Matrix Filling[J].,2020,30(12):200.[doi:10. 3969 / j. issn. 1673-629X. 2020. 11. 037]
[5]张玉瑶,程学林,尹天鹤.基于深度学习和矩阵分解的推荐算法[J].计算机技术与发展,2021,31(07):21.[doi:10. 3969 / j. issn. 1673-629X. 2021. 07. 004]
　ZHANG Yu-yao,CHENG Xue-lin,YIN Tian-he.A Recommendation Algorithm Based on Deep Learning and Matrix Factorization[J].,2021,31(12):21.[doi:10. 3969 / j. issn. 1673-629X. 2021. 07. 004]
[6]任静霞,武志峰.动态信任衰减和信息匹配的混合推荐算法[J].计算机技术与发展,2021,31(10):30.[doi:10. 3969 / j. issn. 1673-629X. 2021. 10. 006]
　REN Jing-xia,WU Zhi-feng.Hybrid Recommendation Algorithm Based on Dynamic Trust Decay and Information Matching[J].,2021,31(12):30.[doi:10. 3969 / j. issn. 1673-629X. 2021. 10. 006]
[7]刘昊东,王 诚.基于热门度修正因子和置信度的协同过滤算法[J].计算机技术与发展,2023,33(03):127.[doi:10. 3969 / j. issn. 1673-629X. 2023. 03. 019]
　LIU Hao-dong,WANG Cheng.Collaborative Filtering Algorithm Based on Popularity Correction Factor and Confidence[J].,2023,33(12):127.[doi:10. 3969 / j. issn. 1673-629X. 2023. 03. 019]
[8]郭英清,王 敏,肖明胜.结合深度知识追踪与矩阵补全的习题推荐方法[J].计算机技术与发展,2023,33(07):188.[doi:10. 3969 / j. issn. 1673-629X. 2023. 07. 028]
　GUO Ying-qing,WANG Min,XIAO Ming-sheng.Recommended Exercise Combining Deep Knowledge Tracking and Matrix Completion[J].,2023,33(12):188.[doi:10. 3969 / j. issn. 1673-629X. 2023. 07. 028]

更新日期/Last Update: 2021-12-10