Research on Context-aware Service Requirement Dynamic Prediction
FENG Kai1，LIU Zhi-zhong2，SONG Cheng1，ZHANG Li1
1（College of Computer Science and Technology，Henan Polytechnic University，Jiaozuo 454003，China）2（School of Computer and Control Engineering，Yantai University，Yantai 264005，China）
Abstract：Service demand forecasting is an important basis for the realization of active service recommendation.How to realize the accurate prediction of service demand has become one of the key problems to be solved in the field of intelligent service.To solve this problem，this paper constructs a deep interaction neural network model with enhanced attention mechanism（Attention Mechanism Enhanced Deep Interaction Network，AMEDIN），and propose a context-aware dynamic forecasting method of service demand based on AMEDIN.Firstly，the method adaptively captures the interaction between different contexts and service requirements through the interaction unit of the AMEDIN model，so as to explicitly model the impact of different contexts on service requirements;then，the impact weights of different contexts on service requirements are obtained based on the attention mechanism by combining context features，interaction relationships and service demand features.Finally，the AMEDIN model is trained based on user features，weighted context features and service requirement features，and the context-aware service demand prediction is realized based on the trained AMEDIN model.A large number of experiments are carried out based on the real datasets provided by Movielens and Alibaba，and the experimental results show that the method proposed in this paper is feasible and effective.
2021, Vol. 42 
