The rapid development of e-commerce has prompted the emergence of massive product reviews. Quickly grasping the emotional tendencies in e-commerce product reviews has become an effective way for businesses to improve their problems. Therefore, based on Bert and GUP, the research proposes Bert BiGRU deep learning algorithm, and verifies its effectiveness in the emotional analysis of e-commerce product reviews. The experimental results show that Bert BiGRU algorithm has the highest accuracy of 95.51% among the seven algorithms. In addition, Bert BiGRU algorithm has achieved good training effect after only eight times of training. In addition, Bert BiGRU algorithm has been used in practice to analyze the emotional tendencies in four mobile phone reviews, which shows high reliability. In a word, Bert BiGRU algorithm has high performance and low cost. It has high practicability in the emotional analysis of e-commerce product reviews.