Human-computer interaction negative emotion analysis method based on bilstm and attention

Abstract

The invention discloses a human-computer interaction negative emotion analysis method based on bilstm and attention. The human-computer interaction negative emotion analysis method comprises the following steps: A, collecting and processing data, and constructing a negative emotion dictionary and a stop dictionary; B, performing word segmentation and Word2vec word vector training on the acquired data; C, constructing a bidirectional long-short-term memory network bilstm, acquiring a word2vec word vector, and inputting the word2vec word vector into the bilstm for context feature extraction; D,adjusting the weight of an attention layer: adding an attention mechanism Attention into a bilstmnetwork, allocating weights to different features in a sentence through the use of the attention mechanism, and paying attention to feature information, which tends to the emotion of a user, in the sentence; and E, outputting an emotion classification result. According to the method, the weight of thekey information can be adjusted, the emotion analysis recognition rate is improved, and finally the man-machine interaction experience is improved.

Description

technical field [0001] The invention relates to the technical field of natural language processing, in particular to a negative emotion analysis method for human-computer interaction based on bilstm and attention. Background technique [0002] Sentiment analysis technology is an important direction in Natural Language Processing (NLP). At present, the main text sentiment analysis researches are mainly three types based on sentiment lexicon-based methods, machine learning-based methods and deep learning-based methods. It relies on sentiment dictionaries and rules, and calculates the sentiment value as the basis for the text's sentimental tendency. The dependence of this kind of method on the sentiment lexicon has become a key factor in its application and development. [0003] main obstacle. Based on the machine learning method, the machine learning method obtains a sentiment analysis classifier by training the manually calibrated data. However, traditional machine learnin...

AI Technical Summary

Problems solved by technology

[0005] After long-term observation and statistics, it is found that in the prior art, in the process of human-computer interaction, when the user does not get a satisfactory answer in multiple interactions due to the colloquialism of the speaker or inaccurate speech recognition, the user's mood will turn bad, and he will say something. uncivilized curse words

Images