融合语义的图神经网络饰品设计知识推荐  

Graph neural network knowledge recommendation for ornaments design fused with semantic

作　　者：刘运通 孙晓莹[2] 张展 LIU Yun-tong;SUN Xiao-ying;ZHANG Zhan(School of Computer and Software,Nanyang Institute of Technology,Nanyang 473004,China;Information Construction and Management Center,Nanyang Institute of Technology,Nanyang 473004,China;School of Computer and Information Engineering,Anyang Normal University,Anyang 455000,China)

机构地区：[1]南阳理工学院计算机与软件学院,河南南阳473004 [2]南阳理工学院信息化建设与管理中心,河南南阳473004 [3]安阳师范学院计算机与信息工程学院,河南安阳455000

出　　处：《计算机工程与设计》2024年第12期3812-3819,共8页Computer Engineering and Design

基　　金：国家自然科学基金项目(62106007);河南省科技攻关基金项目(232102230049);南阳理工学院博士科研启动基金项目(201913)。

摘　　要：为将企业结构化数据中所蕴含的专业知识高效地提供给产品设计师,提出一种基于图注意力网络的饰品设计知识推荐方法。把饰品企业的结构化数据转化为图,依据专业知识图谱融入相关语义信息,构建图注意力网络模型学习设计师的历史设计习惯,用门控循环单元神经网络提取当前设计任务的操作序列特征,依据这些信息,预测饰品设计师所需的知识并进行推荐。实验结果表明,在专业领域知识推荐方面,该方法具有更好的推荐性能。To efficiently provide the knowledge implicated in structured data of enterprises for product designers,a knowledge recommendation method for ornaments design based on graph attention network was proposed.The structured data of ornaments enterprises were converted into graphs and the semantic information obtained from the domain knowledge graph was fused into the graphs,and a graph attention network model was constructed to learn the historical habits of the ornaments designers,and the gated recurrent unit neural(GRU)network was used to extract the features of the designing operation sequence for the current task.According to the information,the knowledge required by the ornaments designers was predicted and the knowledge was recommended to them.Experimental results show that the method has better recommendation performance in professional domain knowledge recommendation.

关 键 词：图神经网络 知识推荐 知识图谱 设计任务子图 注意力机制 门控循环单元 饰品设计 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]