Psychological Analysis
Classifying emotions & Extracting emotions from text
In order to understand the user’s emotion in texts, we had to analyze dialogic utterances in the real world first. There are two basic emotion categories: basic(primary) emotions and second order(derived) emotions. Primary emotions mean ‘pure’ emotional state, such as ‘fear’, ‘anger’, ‘happiness’, ‘sadness’, ‘surprise’, ‘disgust’. Secondary order emotions mean the emotional states are more complicated than basic. Emotions are so extensive and complex that we cannot use all of them or fit the word in each suitable emotion. Therefore, we divided into seven categories of emotions to contain the rest of the emotion words – happiness, anger, sorrow, guilt, embarrassment, shock, and love. Then, we classified a hundred of second order emotion words into those seven categories. Also, we searched for particular words that are used a lot on SNS, podcasts, and messages but not proper emotion terms. They are called “emotion related terms’, which even some of them are not classified as emotion words officially, but still offer the emotional states. Then, we divided those words to 2 sections again – good(positive) and bad(negative). We classified emotions as simple as possible so that we can run programs easier. After all, if the program recognizes one of those words from the texts, it will combine those primary and second order emotion words to finally conclude the user’s emotion. Furthermore, diverse sentence structures will also indicate the user’s emotion, such as the length of the sentence, specific word orders, etc. We are planning to consider sentence structure of the user’s texts later on.
- Tools for describe emotions Naturalistic Database
- Basic Emotion Vocabulary
- Feeltrace
- others
- Pennebaker’s Linguistic Inquiry and Word Count
- Whissell’s Dictionary of Affect in Language
S.Byun