Text sentiment analysis, also known as opinion mining, tendency analysis, etc. In simple terms, it is the process of analyzing, processing, summarizing and inferring subjective texts with emotional colors.
By judging the sentiment tendency of the text containing subjective information, it can support the optimization effect of the online training model and provide help for applications such as word-of-mouth analysis, topic monitoring, and public opinion analysis.
The following content is excerpted from a fine-grained emotional multi-class labeling project.
1. Label classification
Emotions are divided into 3 major categories, 11 middle categories and 26 minor categories:
2. Annotation description
For a short dialogue data, determine the emotion of the speaker in each sentence at that time, and determine whether the sentence contains 26 sub-categories of emotion, if it contains, label the emotion label as 1, and 0 as the contrary.
3. Annotation example
Conversation content:
A: Philip, I'm really happy to hear that you won the award. congratulations!
Annotation example: 
Dialogue content:
B: Thank you. Di Nisi. This actually surprised me. I mean, there are many people who are qualified.
Annotation example: 
Dialogue content:
A: That's true. But your work is indeed excellent, and certainly deserved.
Annotation example: 
Dialogue content:
B: Thank you very much. I look forward to seeing you nominated soon. You have always done a great job.
Annotation example: