摘要
为缓解传统电子产品价值评估方式高度依赖专业人员的干预,需要频繁更新评估算法,人工工作量过大的问题,提出基于属性分类建模的电子产品价值评估方法。通过将产品属性进行分类,分别建模产品属性与价值评估结果的映射关系,引入时间信息来增强预测结果的时效性,采用一种缓解数据驱动模型冷启动问题的方法,获得更加精准的价值评估结果。通过在真实的手机回收订单数据集上进行的实验,验证了该模型在减少人工干预的同时取得了较高的电子产品价值评估精度。
To alleviate the problem that the traditional electronic product value assessment method highly relies on the intervention of professionals,which requires frequent updating the assessment algorithm and leads to excessive manual workload,an electronic product value assessment method based on attribute classification modeling was proposed,which modeled the mapping relationship between product attributes and value assessment results separately by classifying product attributes.Time information was introduced to enhance the timeliness of the prediction results.A method was adopted to alleviate the cold start problem of the data-driven model and more accurate value assessment results were obtained.The model was implemented on a real cell phone recycling order dataset.Results demonstrate that the model achieves high accuracy of electronic product value evaluation with less human intervention.
作者
王陆霖
刘杨
彭治
杜永萍
韩红桂
WANG Lu-lin;LIU Yang;PENG Zhi;DU Yong-ping;HAN Hong-gui(Faculty of Information Technology,Beijing University of Technology,Beijing 100124,China)
出处
《计算机工程与设计》
北大核心
2022年第7期2040-2047,共8页
Computer Engineering and Design
基金
国家重点研发计划基金项目(2018YFC1900804)。
关键词
价值评估
电子产品回收
属性分类
神经网络
主成分分析
value evaluation
electronics recycling
attribute classification
neural network
principal component analysis