Multi-label prediction of cigarettes based on odor knowledge graphs
In the digital economy,there is a consensus to use big data and artificial intelligence technologies to help transform and upgrade cigarette retail management.Label prediction effectively matches cigarette products and consumers'needs in the new retail model to promote sales.In this paper,we propose to combine knowledge graph and natural language processing technology to enhance consumer perception wild from different dimensional features in the new retail model,and use odor knowledge graph to drive each scent sensory and data view to jointly learn the contribution of multi-label features,maintain the consistency between feature kernels and the similarity of poten-tial representations,and realize multi-label prediction of finished cigarettes,so as to improve cigarette sales effi-ciency and enhance consumer experience perception.The experimental results demonstrate the effectiveness of this approach in the field of cigarette marketing management prediction.
odor knowledge graphnatural language processinglabel predictionattention mechanism
国家科技期刊平台
NSTL
万方数据
