本研究旨在通过文本挖掘方法研究消费者的需求和偏好。通过收集和预处理天猫商城的服装商品的在线评论数据,应用BERT-LDA模型进行分析,发现消费者在购物体验、服装特性和服装品质方面呈现出多样化的关注度和情感积极率。研究结果表明,...本研究旨在通过文本挖掘方法研究消费者的需求和偏好。通过收集和预处理天猫商城的服装商品的在线评论数据,应用BERT-LDA模型进行分析,发现消费者在购物体验、服装特性和服装品质方面呈现出多样化的关注度和情感积极率。研究结果表明,虚拟试穿等新型产品体验方式将深刻影响消费者的购买决策。消费者提高了对服装的可持续性的关注程度,倾向于选择实用性强、易于回收利用,且能“一衣多穿”的服装。基于该研究结果,本文为服装电商行业的市场营销提供了有益的参考和指导。The purpose of this study is to study consumers’ needs and preferences through text mining methods. By collecting and preprocessing online review data of clothing products on Tmall and applying BERT-LDA model for analysis, it is found that consumers show diversified attention and positive emotional rate in terms of shopping experience, clothing characteristics and clothing quality. The results show that new product experience methods such as virtual trying on will profoundly affect consumers’ purchasing decisions. Consumers are paying more attention to the sustainability of clothing, and tend to choose clothes that are practical, easy to recycle, and can be worn more than once. Based on the research results, this paper provides useful reference and guidance for the marketing of apparel e-commerce industry.展开更多
传统的主题模型在对短文本建模时,会由于词汇稀疏导致模型效果不好。论文针对短文本数据特征,提出了BERT-LDA主题挖掘模型(Short Text Topic Mining Based on BERT and LDA),该算法通过使用预训练BERT模型提取文本语义特征,再通过K-mean...传统的主题模型在对短文本建模时,会由于词汇稀疏导致模型效果不好。论文针对短文本数据特征,提出了BERT-LDA主题挖掘模型(Short Text Topic Mining Based on BERT and LDA),该算法通过使用预训练BERT模型提取文本语义特征,再通过K-means聚类算法将短文本聚合成长文本再进行主题建模,从而扩充单条文本包含的语义特征,有效降低了词汇稀疏性,从而提升模型效果。通过在实际数据上进行对比实验证明,与LDA和BTM模型相比,该算法能够取得更低的困惑度。展开更多
文摘本研究旨在通过文本挖掘方法研究消费者的需求和偏好。通过收集和预处理天猫商城的服装商品的在线评论数据,应用BERT-LDA模型进行分析,发现消费者在购物体验、服装特性和服装品质方面呈现出多样化的关注度和情感积极率。研究结果表明,虚拟试穿等新型产品体验方式将深刻影响消费者的购买决策。消费者提高了对服装的可持续性的关注程度,倾向于选择实用性强、易于回收利用,且能“一衣多穿”的服装。基于该研究结果,本文为服装电商行业的市场营销提供了有益的参考和指导。The purpose of this study is to study consumers’ needs and preferences through text mining methods. By collecting and preprocessing online review data of clothing products on Tmall and applying BERT-LDA model for analysis, it is found that consumers show diversified attention and positive emotional rate in terms of shopping experience, clothing characteristics and clothing quality. The results show that new product experience methods such as virtual trying on will profoundly affect consumers’ purchasing decisions. Consumers are paying more attention to the sustainability of clothing, and tend to choose clothes that are practical, easy to recycle, and can be worn more than once. Based on the research results, this paper provides useful reference and guidance for the marketing of apparel e-commerce industry.
文摘传统的主题模型在对短文本建模时,会由于词汇稀疏导致模型效果不好。论文针对短文本数据特征,提出了BERT-LDA主题挖掘模型(Short Text Topic Mining Based on BERT and LDA),该算法通过使用预训练BERT模型提取文本语义特征,再通过K-means聚类算法将短文本聚合成长文本再进行主题建模,从而扩充单条文本包含的语义特征,有效降低了词汇稀疏性,从而提升模型效果。通过在实际数据上进行对比实验证明,与LDA和BTM模型相比,该算法能够取得更低的困惑度。
文摘构建大规模网络舆情演化仿真模型,对新冠疫情武汉重灾区与全国其他地区采取差异化的应急管理和舆情疏导具有指导价值。为实现主题细粒度的舆情情感演化仿真,将LDA(Latent Dirichlet Allocation)主题模型与BERT(Bidirectional Encoder Representations from Transformers)词向量深度融合,优化主题向量助力文本主题聚类;同时,在改进BERT预训练任务的基础上,叠加深度预训练任务,以提高模型在情感分类中的精确度。结果表明:在主题向量训练过程中,改进的BERT-LDA模型较原始LDA模型NPMI(Normalized Pointwise Mutual Information)值提升0.357;在疫情事件情感分类任务上,AUC(Area Under the Curve)值超过了99.6%,证明其能够有效运用于大规模网络舆情演化仿真。
