A social review sentiment explainable identification method and system

Abstract

The application discloses a social comment emotion explainable identification method and system, acquires social comment data and processes; uses a support vector machine (SVM) algorithm to train a preprocessed data set, obtains a trained SVM model; obtains support vectors after the SVM model training is completed and views, identifies key feature words for emotion judgment, and constructs a dynamic key feature word library; a double-branch combined network (DCN) is established, which is used for carrying out sentiment analysis on social media short comments according to the identified key feature words, and realizing word-level and sentence-level explainability. The application discloses a correlation model output and a feature word and a key word sequence, reveals the logic and basis of emotion recognition, and thus improves the credibility and transparency of the model.

Images