Incremental Recommendation Algorithms Based on Word Embedding Model and Neural Networks
Xuemei Zhang, Lei Shi · 2023
Traditional collaborative filtering recommender systems need to retrain the entire dataset in case of drastic data changes, which generates huge computational overhead in big data scenarios. In view of this,an incremental collaborative filtering recommendation system based on sequence similarity measurement and neural networks was proposed. In the offline preprocessing phase, the user profile, explicit feedback information and implicit feedback information were modeled as sequence formatted contexts, and a sequence-based similarity metric is designed; the word embedding model and the collaborative filtering system are combined to predict the list of items favored by the user using the context. This result for that experiment reflects that the recommendation results of the algorithm exhibit better alignment satisfaction, novelty and diversity, and it supports incremental system update processing as well.