Emotion classification method based on multi-source information fusion

The invention discloses an emotion classification method based on multi-source information fusion. The method comprises the steps: S1, selecting part of comment data of a target user in a specified Internet platform, setting an emotion label, carrying out the label propagation according to the emotion similarity relation of the comment data with the same or similar score, and obtaining the expanded label data; S2, extracting dictionary sentiment information based on the plurality of sentiment dictionaries, and extracting sentiment information including sentiment word co-occurrence information,sentiment word polarity information and sentiment word pair polarity relation information to obtain multi-source sentiment information; S3, after the emotion dictionary is initialized, continuously iterating and updating the emotion dictionary to continuously approach the extracted multi-source emotion information, and constructing a required emotion dictionary; and S4, performing emotion classification on the comment da