摘要
目的 探索基于机器学习的开放创新创意识别方法,解决创意识别过程中存在的耗时长、效率低、成本高等问题。方法 从用户特征、用户参与度和创意内容特征三个方面构建评估模型,以OpenIDEO社区为研究对象,采集数据并进行数据清洗和数据转化映射,最后进行多种机器学习算法的参数优化,并以F1值为选择标准,选择分类效果最佳的算法作为分类模型。结果 运用KNN、SVM、决策树、随机森林四种机器学习算法分析OpenIDEO数据,随机森林算法通过参数优化取得了最大的F1值(0.919 09),同时对于验证数据,该算法同样可以取得较好的分类效果。结论 应用机器学习方法对开放式创新社区中的创意进行识别,具有较高的可行性和有效性,可以大大降低社区在创意筛选中的投入,提高创新效率,优化社区生态。
The work aims to explore a machine learning-based method of creative identification in open innovation to solve a number of problems in creative identification,including long time consumption,low efficiency and high costs.An evaluation model was built at first in light of user characteristics,degree of user participation and creative content char-acteristics,followed by focusing research on OpenIDEO Community,collecting data,conducting data cleansing and data conversion mapping.Finally,parameter optimization of multiple machine learning algorithms was made with F1 value as selection criterion to select the algorithm with the best classification effect as classification model.Four machine learning algorithms,including KNN,SVM,decision tree and random forest,were adopted to analyze OpenIDEO data,among which the random forest algorithm generated the maximum F1 value(0.91909)through parameter optimization.For data validation,such algorithm could also produce good classification effects.The adoption of machine learning method is highly feasible and effective for creative identification in open innovation community.It can greatly reduce community's input in creative screening,improve innovation efficiency and optimize community ecology.
作者
薛水晶
郭伟
连文涛
张静
XUE Shui-jing;GUO Wei;LIAN Wen-tao;ZHANG Jing(Tianjin Key Laboratory of Equipment Design and Manufacturing Technology,Tianjin University,Tianjin 300350,China;Tianjin Ren'ai College,Tianjin 301636,China)
出处
《包装工程》
CAS
北大核心
2023年第8期72-80,共9页
Packaging Engineering
基金
天津市教委科研计划项目(2021KJ082)。
关键词
开放创新
创新社区
创意识别
机器学习
open innovation
innovation community
creative identification
machine learning
