System, method and computer program product for identifying words within collection of text applicable to specific sentiment
A content intelligence module may implement a sentiment analysis method to identify words or phrases from user-generated content that are associated with a particular sentiment. The method may comprise grouping or splitting text into different sentiment segments, tokenizing words or phrases and/or removing stopwords across the sentiment segments, performing a frequency analysis to count the words or phrases in each sentiment segment, scaling the frequency results across the sentiment segments where necessary, and removing commonly used words from the sentiment segments. The words or phrases that are left in a specific sentiment segment are the most-used words for that sentiment segment. The word cloud module therefore allows for very quick generation of a summary around sentiment segments. A sentiment overview containing the summary can be presented to a user in connection with a selected product or service with which the user-generated content is associated.
1. A method for analyzing sentiment, comprising:
at a first computer:
dividing a collection of text into a plurality of sentiment segments;
tokenizing words or phrases in the plurality of sentiment segments;
performing a frequency analysis on tokenized words or phrases in each sentiment segment of the plurality of sentiment segments;
performing a scaling operation to size individual sentiment segments based on results from the frequency analysis;
for each tokenized word or phrase in each sentiment segment of the plurality of sentiment segments, subtracting a first number of the tokenized word or phrase in the sentiment segment from a second number of the tokenized word or phrase in at least one other sentiment segment of the plurality of sentiment segments, thereby producing, for each sentiment segment of the plurality of sentiment segments, a list of words or phrases that apply specifically to the sentiment segment; and
providing the list of words or phrases that apply specifically to the sentiment segment to a second computer over a network connection.
2. The method of claim 1 , further comprising:
removing stopwords in each sentiment segment of the plurality of sentiment segments prior to performing the frequency analysis.
3. The method of claim 1 , wherein the collection of text is divided into the plurality of sentiment segments based on structured information about the text.
4. The method of claim 1 , wherein the collection of text comprises user-generated content.
5. The method of claim 1 , wherein the collection of text is associated with a product or service.
6. The method of claim 1 , wherein the scaling operation is performed based on an overall count of the tokenized words or phrases in each sentiment segment of the plurality of sentiment segments.
7. The method of claim 1 , wherein the plurality of sentiment segments comprises a positive sentiment and a negative sentiment.
8. A computer program product comprising at least one non-transitory computer readable medium storing instructions translatable by a first computer to:
divide a collection of text into a plurality of sentiment segments;
tokenize words or phrases in the plurality of sentiment segments;
perform a frequency analysis on tokenized words or phrases in each sentiment segment of the plurality of sentiment segments;
perform a scaling operation to size individual sentiment segments based on results from the frequency analysis;
for each tokenized word or phrase in each sentiment segment of the plurality of sentiment segments, subtract a first number of the tokenized word or phrase in the sentiment segment from a second number of the tokenized word or phrase in at least one other sentiment segment of the plurality of sentiment segments, thereby producing, for each sentiment segment of the plurality of sentiment segments, a list of words or phrases that apply specifically to the sentiment segment; and
provide the list of words or phrases that apply specifically to the sentiment segment to a second computer over a network connection.
9. The computer program product of claim 8 , wherein the instructions are further translatable by the first computer to perform:
removing stopwords in each sentiment segment of the plurality of sentiment segments prior to performing the frequency analysis.
10. The computer program product of claim 8 , wherein the collection of text is divided into the plurality of sentiment segments based on structured information about the text.
11. The computer program product of claim 8 , wherein the collection of text comprises user-generated content.
12. The computer program product of claim 8 , wherein the collection of text is associated with a product or service.
13. The computer program product of claim 8 , wherein the scaling operation is performed based on an overall count of the tokenized words or phrases in each sentiment segment of the plurality of sentiment segments.
14. A system, comprising:
at least one processor;
at least one non-transitory computer readable medium storing instructions translatable by the at least one processor to implement a word cloud module, the word cloud module being configured to:
divide a collection of text into a plurality of sentiment segments;
tokenize words or phrases in the plurality of sentiment segments;
perform a frequency analysis on tokenized words or phrases in each sentiment segment of the plurality of sentiment segments;
perform a scaling operation to size individual sentiment segments based on results from the frequency analysis;
for each tokenized word or phrase in each sentiment segment of the plurality of sentiment segments, subtract a first number of the tokenized word or phrase in the sentiment segment from a second number of the tokenized word or phrase in at least one other sentiment segment of the plurality of sentiment segments, thereby producing, for each sentiment segment of the plurality of sentiment segments, a list of words or phrases that apply specifically to the sentiment segment; and
provide the list of words or phrases that apply specifically to the sentiment segment to a second computer over a network connection.
15. The system of claim 14 , wherein the word cloud module is further configured to perform:
removing stopwords in each sentiment segment of the plurality of sentiment segments prior to performing the frequency analysis.
16. The system of claim 14 , wherein the collection of text is divided into the plurality of sentiment segments based on structured information about the text.
17. The system of claim 14 , wherein the collection of text comprises user-generated content.
18. The system of claim 14 , wherein the collection of text is associated with a product or service.
19. The system of claim 14 , wherein the scaling operation is performed based on an overall count of the tokenized words or phrases in each sentiment segment of the plurality of sentiment segments.
20. The system of claim 14 , wherein the plurality of sentiment segments comprises a positive sentiment and a negative sentiment.