基于深度学习的商品推荐系统研究

北京理工大学珠海学院2020届本科生毕业论文Research on Commodity Recommendation System Based on DeepLearningAbstractIn order to more quickly and accurately recommend products of interest to users,this paperstudies the recommendation system from three directions such as traditional algorithm anddata mining and deep leaming.Alibaba's product behavior data of 8477 users from Novem-ber 16 to December 16,2018 are selected.Feature extraction is based on two dimensionsof users and goods.And we use R and Python to implement the recommendation systemfor LR and GBDT and DeepFM respectively.Finally,we conclude that DeepFM has betterprediction effect than LR and GBDT.It can more accurately judge whether to recommend ornot for users to buy this product.It also gives some suggestions to the merchants.Throughthe prediction of the recommendation system,it can promote the users who are interestedin the products,so as to increase the income.Keywords:Recommendation System DeepFM Algorithm GBDT Algorithm LR Al-gorithm