基于多维用户画像和DeepFM的“环评云助手”资源推荐研究

基于多维用户画像和DeepFM的“环评云助手”资源推荐研究.pdf

摘要(Abstract)：

“环评云助手”是一款服务于环境影响评价行业用户的APP,针对APP中信息量激增、行业资源文本特征利用不充分和行业用户即时资源推荐精准较低等问题，提出一种结合行业文本资源和用户行为特征的多维用户画像模型并应用于深度因子分解机(Deep Factorization Machines, DeepFM),实现资源点击率(Click-Through-Rate, CTR)的精准预测.模型首先对行业资源文本进行语义抽取，再对行业用户行为进行自定义评分，从而构建多维用户画像模型；最后将多维用户画像应用于DeepFM模型，进行CTR预测任务，实现具有行业特征的个性化推荐.实验数据来自“环评云助手”APP,实验结果表明该模型有效提高了CTR预测任务的AUC值，降低了LogLoss值，具有一定的应用价值.
EIA Cloud is an APP that serves users in the Environmental Impact Assessment. In view of the problems such as the surge of information, the insufficient use of text features and the low accuracy of real-time resource recommendation by users, the paper proposes a multi-dimensional user portrait model based on DeepFM combined with industry resources and user behavior to achieve CTR prediction. Firstly, the industry resource semantics is extracted, and then the user behavior is scored to build a multi-dimensional user portrait model. Finally, the model is applied to DeepFM to perform CTR prediction and achieve personalized recommendation with industry feature. Experimental data are obtained from EIA Cloud. The experimental results show that the model can effectively improve the AUC value of CTR prediction tasks and reduce the LogLoss value, which has certain application value.

关键词(KeyWords)：环境影响评价;用户画像;标签生成;推荐算法;深度因子分解机;CTR预测任务
Environmental Impact Assessment;user portrait;tag generation;recommendation algorithm;DeepFM;CTR prediction

基金项目(Foundation):国家自然科学基金(51975058);; 教育部人文社科规划基金(20YJAZH129);; 2022年北京信息科技大学优质课程专项;; 北京信息科技大学2023课程思政立项项目(2023JGSZ20)

作者(Authors):李天玉;车蕾;丁峰;谭悦;
Li Tianyu;Che Lei;Ding Feng;Tan Yue;School of Information Management, Beijing Information Science and Technology University;Beijing Shangyun Co., LTD.;

DOI:10.16366/j.cnki.1000-2367.2023.04.006

参考文献(References)：

• [1] TEIXEIRA C,PINTO J,MARTINS J.User profiles in organizational environments[J].Campus-Wide Information Systems,2008,25(3):329-332.

• [2] QUINTANA R M,HALEY S R,LEVICK A,et al.The Persona party:using personas to design for learning at scale[C]//Proceedings of the 2017 CHI Conference Extended Abstracts on Human Factors in Computing Systems.New York:ACM,2017:933-941.

• [3] 刘海鸥，黄文娜，姚苏梅，等.基于深度学习的移动图书馆用户画像情境化推荐[J].图书馆学研究，2019(21):57-64.LIU H O,HUANG W N,YAO S M,et al.Situation recommendation for mobile library users' portrait based on deep learning[J].Research on Library Science,2019(21):57-64.

• [4] 赵辉，化柏林，何鸿魏.科技情报用户画像标签生成与推荐[J].情报学报，2020,39(11):1214-1222.ZHAO H,HUA B L,HE H W.User profile tag generation and information recommendations for science and tencnology intelligence[J].Journal of the China Society for Scientific and Technical Information,2020,39(11):1214-1222.

• [5] GOLDBERG D,NICHOLS D,OKI B M,et al.Using collaborative filtering to weave an information tapestry[J].Communications of the ACM,1992,35(12):61-70.

• [6] 张亮.基于LDA主题模型的标签推荐方法研究[J].现代情报，2016,36(2):53-56.ZHANG L.Research on tagging recommendation method based on LDA topic model[J].Journal of Modern Information,2016,36(2):53-56.

• [7] 熊回香，窦燕.基于LDA主题模型的标签混合推荐研究[J].图书情报工作，2018,62(3):104-113.XIONG H X,DOU Y.Research on tag hybrid recommendation based on LDA topic model[J].Library and Information Service,2018,62(3):104-113.

• [8] 熊回香，杨雪萍，高连花.基于用户兴趣主题模型的个性化推荐研究[J].情报学报，2017,36(9):916-929.XIONG H X,YANG X P,GAO L H.Personalized recommendation research based on user interest topic model[J].Journal of the China Society for Scientific and Technical Information,2017,36(9):916-929.

• [9] 熊回香，杨雪萍.社会化标注系统中的个性化信息推荐研究[J].情报学报，2016,35(5):549-560.XIONG H X,YANG X P.Personalized information recommendation research based on combined condition in folksonomies[J].Journal of the China Society for Scientific and Technical Information,2016,35(5):549-560.

• [10] 李兴华，陈冬林，杨爱民，等.基于用户兴趣-标签的混合推荐方法研究[J].情报学报，2015,34(5):466-470.LI X H,CHEN D L,YANG A M,et al.A study of mixed recommendation method based on user interest-tag[J].Journal of the China Society for Scientific and Technical Information,2015,34(5):466-470.

• [11] JUAN Y,ZHUANG Y,CHIN W S,et al.Field-aware factorization machines for CTR prediction[C]//Proceedings of the 10th ACM Conference on Recommender Systems.New York:ACM,2016:43-50.

• [12] GUO H F,TANG R M,YE Y M,et al.DeepFM:a factorization-machine based neural network for CTR prediction[C]//Proceedings of the 26th International Joint Conference on Artificial Intelligence.New York:ACM,2017:1725-1731.

• [13] 李维，闫晓东，解晓庆.基于改进TextRank的藏文抽取式摘要生成[J].中文信息学报，2020,34(9):36-43.LI W,YAN X D,XIE X Q.An improved TextRank for Tibetan summarization[J].Journal of Chinese Information Processing,2020,34(9):36-43.

• [14] 车蕾，杨小平.多特征融合文本聚类的新闻话题发现模型[J].国防科技大学学报，2017,39(3):85-90.CHE L,YANG X P.News topic discovery model of multi feature fusion text clustering[J].Journal of National University of Defense Technology,2017,39(3):85-90.

• [15] 蒋兴渝，黄贤英，陈雨晶，等.特征重要性动态提取的广告点击率预测模型[J].小型微型计算机系统，2022,43(5):976-984.JIANG X Y,HUANG X Y,CHEN Y J,et al.Advertising click-through rate prediction model based on dynamic extraction of feature importance[J].Journal of Chinese Computer Systems,2022,43(5):976-984.

• [16] 付芬，豆育升，韩鹏，等.基于隐式评分和相似度传递的学习资源推荐[J].计算机应用研究，2017,34(12):3725-3729.FU F,DOU Y S,HAN P,et al.Learning resource recommendation based on implicit scoring and similarity propagation[J].Application Research of Computers,2017,34(12):3725-3729.

• [17] 顾寰，杨长春，吴云，等.融合社区结构和个人兴趣的协同过滤推荐算法[J].计算机工程与设计，2018,39(11):3420-3424.GU H,YANG C C,WU Y,et al.Collaborative filtering recommendation algorithm combining community structure and personal interests[J].Computer Engineering and Design,2018,39(11):3420-3424.

• [18] LIAN J X,ZHOU X H,ZHANG F Z,et al.xDeepFM:combining explicit and implicit feature interactions for recommender systems[C]//Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.New York:ACM,2018:1754-1763.