Social network text sentiment analysis method based on deformable self-attention mechanism

Abstract

The invention discloses a social network text sentiment analysis method based on a deformable self-attention mechanism, which is used for analyzing the sentiment of a user utterance. The method comprises the following steps: segmenting each sentence in user utterance text data into words, and expressing each word with a word vector; inputting the word vector sequence into a bidirectional recurrentneural network (Bi-LSTM) to obtain a coded representation of each word; utilizing a deformable self-attention mechanism to convert the coding representation of the words into a plurality of sentencecoding representations with different context ranges; fusing the plurality of sentence code representations to obtain a sentence code representation; inputting the fused sentence code representation into a feed-forward neural network (FFN) for classification, and outputting a result; according to a model output result and a data real result, iteratively training a model updating parameter by minimizing a cross entropy loss function; and inputting the social network text to be classified into the trained model to obtain an emotion analysis result.

Description

technical field [0001] The invention relates to the technical field of social network text sentiment analysis in natural language processing, in particular to a social network text sentiment analysis method based on a deformable self-attention mechanism. Background technique [0002] A platform for content production and exchange based on user relationships in the Internet is a social network, where people can share opinions with each other. Using the text sentiment analysis method, it is possible to automatically extract the user's emotional preference for some things such as products, services, events, etc., so as to help users better choose their favorite products and help businesses provide better products and services. The use of public opinion and sentiment analysis can also predict the public's attitude towards social events and the trend of emotional changes, which is conducive to enterprises and government agencies to make corresponding adjustments in real time. Se...

AI Technical Summary

Problems solved by technology

Some people use the self-attention model (see: Z.Lin, M.Feng, C.N.dos Santos, M.Yu, B.Xiang, B.Zhou, and Y.Bengio, "Astructured self-attentive sentence embedding," 2017.), however, because RNN encodes each word sequentially, it does not directly model multi-scale contextual features, and the self-attention model extracts global contextual features

Some local self-attention models also consider local context features, see: T.Shen, T.Zhou, G.Long, J.Jiang, and C.Zhang, "Tensorized self-attention: Efficiently modeling pairwise and global dependencies together ," in Proc.Conf.North Amer.ChapterAssoc.Comput.Linguistics, 2019, pp.1256–1266.), but the context features extracted by this model are also fixed-scale

At present, the methods at home and abroad are not very good at learning local context features of different scales for different words, and all of them are fixed-scale or global-scale contexts.

Images