Method and apparatus for analyzing affect and emotion in text
First Claim
Patent Images
1. A computer-assisted method for classifying a text document according to emotion and affect, comprising the steps of:
- assigning a score to each affect term in the document, computing an affect score for the document from the scores for each affect term, classifying the document in accordance with the affect score, classifying affect terms as positive or negative, identifying named entities in the document, assigning the scores of the affect terms to the named entities, and for each named entity, summing the scores of all positive affect terms, summing the scores of all negative affect terms, and computing an affect score by subtracting the sum of the scores of all negative affect terms from the sum of the scores of all positive affect terms.
1 Assignment
0 Petitions
Accused Products
Abstract
Disclosed is a computer-assisted method for classifying a text document according to emotion and affect. A score is assigned to each affect term in the document. An affect score is computed for the document from the scores for each affect term. The document is then classified in accordance with the affect score. An apparatus for performing the computer-assisted method is also disclosed.
191 Citations
20 Claims
-
1. A computer-assisted method for classifying a text document according to emotion and affect, comprising the steps of:
-
assigning a score to each affect term in the document, computing an affect score for the document from the scores for each affect term, classifying the document in accordance with the affect score, classifying affect terms as positive or negative, identifying named entities in the document, assigning the scores of the affect terms to the named entities, and for each named entity, summing the scores of all positive affect terms, summing the scores of all negative affect terms, and computing an affect score by subtracting the sum of the scores of all negative affect terms from the sum of the scores of all positive affect terms. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19)
assigning an intensity value for each term, and counting the number of times each affect term occurs within the document, wherein the score assigned to each affect term is the number of times the term occurs multiplied by the intensity value for the terms as the score.
-
-
5. The computer-assisted method according to claim 1, wherein if a negation term is used in conjunction with the affect term, the positive affect term is treated as a negative affect term, and vice versa.
-
6. The computer-assisted method according to claim 1, wherein each affect term is classified as positive or negative at each occurrence of the term based on the part of speech of the affect term at the occurrence.
-
7. The computer-assisted method according to claim 1, wherein each affect term is classified as positive or negative at each occurrence of the affect term based on the meaning of the affect term at the occurrence.
-
8. The computer-assisted method according to claim 1, wherein the affect score is the sum of the scores of all positive affect terms and the sum of the scores of all negative affect terms.
-
9. The computer-assisted method according to claim 1, wherein the score assigned to each affect term is assigned by additionally parsing sentences of the document such that each verb'"'"'s score is assigned to the verb'"'"'s agent and objects and each modifier'"'"'s score is assigned to the modifier'"'"'s objects.
-
10. The computer-assisted method according to claim 1, wherein the classification of the affect term as positive or negative is based on the direction from the affect term to the named entity.
-
11. The computer-assisted method according to claim 1, wherein the assignment of the affect term score to the named entity occurs when the affect term and the named entity are in the same sentence.
-
12. The computer-assisted method according to claim 1, wherein the scores from the affect terms are assigned to the named entities by, for each affect term, assigning the score from the affect term to the closest named entity.
-
13. The computer-assisted method according to claim 1, wherein the scores from the affect terms are assigned to the named entities by, for each named entity, assigning the score from the closest affect term to the named entity.
-
14. The computer-assisted method according to claim 1, wherein the assignment occurs when no other named entity is between the named entity and the closest affect term.
-
15. The computer-assisted method according to claim 1, further comprising the step of:
canonicalizing variants of the named entities into groups of synonymous variants to be treated as a single named entity.
-
16. The computer-assisted method according to claim 1, wherein the text document is a news article.
-
17. The computer-assisted method according to claim 16, wherein the news article is a financial news article.
-
18. The computer-assisted method according to claim 1, wherein the text document is a web page.
-
19. The computer-assisted method according to claim 1, wherein the text document is customer communications.
-
20. An apparatus to enable a method for classifying a text document according to emotion and affect, comprising:
-
means for assigning a score to each affect term in the document, means for computing an affect score for the document from the scores for each affect term, means for classifying the document in accordance with the affect score, means for classifying the affect terms as positive or negative, means for identifying named entities in the document, means for assigning the scores of the affect terms to the named entities, and for each named entity, means for summing the scores of all positive affect terms, means for summing the scores of all negative affect terms, and means for computing an affect score by subtracting the sum of the scores of all negative affect terms from the sum of the scores of all positive affect terms.
-
Specification