Using Incremental Algorithm in Hybrid Recommender System Combined Sentiment Analysis

Abstract

Beside the problem how to improve hybrid system combine sentiment analysis, developing incremental algorithms become an interesting research in real-data en-vironment. While improving the extension of Vietnamese language sentiment analysis is still difficult, stochastic gradient descent algorithm (SGD) exposes the limitations about optimal process in incremental learning. Sterm from two issues, the study proposed model combine Long Short Term Memory with KSGD algorithms in matrix factorization to improve the time and accuracy of predict model. With experimental results, this work proves that proposed system achieves better results with accuracy and learning time.