Probabilistic reasoning and emotion cognition-based text fine-grained emotion generation method

Abstract

The invention relates to a probabilistic reasoning and emotion cognition-based text fine-grained emotion generation method. The method comprises the following steps of 1: preparing a text data set required by a training method; 2: processing the text data set; 3: extracting an emotion assessment variable used for constructing a Bayesian network; 4: according to characteristics of a network text, adding an expression symbol and word frequency-based emotion assessment variable; 5: constructing an emotion knowledge base; 6: constructing a commonsense knowledge base; 7: performing value assignmenton the emotion assessment variable; 8: performing emotion learning to generate a network structure of the Bayesian network; 9: performing parameter learning; and 10: finishing construction work of the emotion generation method. The problem that hidden emotions are ignored in other emotion generation methods is solved by utilizing an emotion cognition method; the emotion generation probability iscalculated by utilizing the Bayesian network; the emotion type probabilities are compared; and one or multiple emotions comprised in the text are generated.

Description

technical field [0001] The present invention relates to the field of machine learning and natural language processing text emotion analysis, specifically relates to emotion cognition methods and the field of reasoning thought of Bayesian network, especially relates to text fine-grained emotion generation method based on probabilistic reasoning and emotion cognition. Background technique [0002] Text sentiment analysis is a popular research field in natural language processing, mainly to automatically identify the emotional categories expressed by users in text. Emotion is a psychological and physiological state produced by a variety of feelings, thoughts and behaviors. It generally refers to like, dislike, anger, surprise, pride, etc. Traditional sentiment analysis is mainly to identify the user's emotional tendency: positive, negative or neutral, while sentiment analysis is a fine-grained sentiment analysis task that can identify multiple emotions. Emotional generation me...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the defects of the prior art, the present invention proposes a text fine-grained emotion generation method based on probabilistic reasoning and emotion cognition, using the emotion cognition method to solve the problem of ignoring implicit emotions in other emotion generation methods, and using Bayeux The Si network calculates the probability of emotion generation and improves the accuracy of emotion category generation

Images