Research on Post Earthquake Public Opinion Analysis Based on XLNet-BiGRU-A Algorithm

Abstract

Because the traditional public opinion sentiment analysis can only be one-way analysis according to the direction of semantic information, it is impossible to fully obtain the semantics of the above and below and extract the deep semantic information in the sentence; and the public opinion comments after the earthquake are different from the general e-commerce, movie and other comments with the characteristics of short sentences and few extractable features. Aiming at the above problems, this paper proposes an emotional trend analysis model that combines XLNet, BiGRU and attention mechanism (XLNetBiGRU-A).The results show that on the self-dataset, the XLNetBiGRU-AT public opinion sentiment trend analysis model has certain advantages, and the accuracy rate is 79.96%. It was increased to 90.42%.