Method for analyzing sentiment trends based on term taxonomies of user generated content
A method for generating a trend report for a non-sentiment phrase. The method comprises generating a plurality of term taxonomies between non-sentiment phrases and sentiment phrases; storing the plurality of term taxonomies in a database; performing periodically at least a statistical analysis respective of the plurality of term taxonomies in the database; receiving a request for a report with respect of at least a non-sentiment phrase in the database; generating a trend report based at least on the at least statistical analysis; and providing the report to the requestor of the report.
1. A system for analyzing sentiment trends based on term taxonomies of user generated content, comprising:
a network interface enabling access to one or more data sources through a network;
a mining unit for collecting textual content from the one or more data sources and generating phrases, the phrases including sentiment phrases and non-sentiment phrases, each of the sentiment phrases including one or more words describing a sentiment, the sentiment being any one of a positive sentiment, a neutral sentiment, and a negative sentiment, each of the sentiment phrases being associated with a score indicating whether the sentiment phrase describes the sentiment;
a non-transitory data warehouse storage medium connected to the network; and
an analysis unit for generating term taxonomies that include at least associations between a non-sentiment phrase and at least one sentiment phrase based on the generated phrases, wherein generated term taxonomies are saved in the data warehouse,
the analysis unit is further configured to periodically statistically analyze the stored term taxonomies to determine at least a statistical trend by performing a correlation analysis on at least two term taxonomies including performing a first correlation and a second correlation, wherein the first correlation is performed on a first stored term taxonomy comprising a non-sentiment phrase and a first sentiment phrase provided by a first group of users and wherein the second correlation is performed on a second stored term taxonomy including said non-sentiment phrase and a second sentiment phrase provided by a second group of users, wherein the first sentiment phrase is different from the second sentiment phrase and the first group of users is different from the second group of users, the analysis unit is further configured to store results of the at least statistical trend in the data warehouse storage, wherein the stored results are available upon receiving a request to generate a trend report regarding the at least statistical trend, wherein the analysis unit is further configured to perform a dynamic weighting of trends from different data sources of the one or more data sources based on at least a frequency that taxonomies results from each data source change.
2. The system of claim 1 , wherein the analysis unit is further configured to provide the trend report respective of at least one sentiment phrase with respect of at least one non-sentiment phrase.
3. The system of claim 1 , wherein the system further comprises:
an alert subsystem configured to generate an alert responsive of any one of: a trend detected from the trend report, crossing of a threshold determined from the trend report, a change in the trend, and a change in frequency occurrence of term taxonomies.
4. The system of claim 1 , wherein the data source is at least one of: a social network, a blog, a web page, a news feed.
5. The system of claim 1 , wherein the analysis unit further provides a recommendation to increase or decrease advertisement expenditure.
6. The system of claim 5 , wherein advertisement expenditure is increased or decreased based on the correlation analysis between the at least two term taxonomies.
7. An apparatus for trend analysis of non-sentiment phrases appearing in at least one data source, comprising:
a mining unit for periodically associating one or more sentiment phrases to one or more non-sentiment phrases and generating a plurality of term taxonomies by associating between the one or more non-sentiment phrases and the one or more sentiment phrases, wherein each of the one or more sentiment phrases includes one or more words describing a sentiment, the sentiment being any one of a positive sentiment, a neutral sentiment, and a negative sentiment, each of the one or more sentiment phrases being associated with a score indicating whether the sentiment phrase describes the sentiment;
a non-transitory storage unit for storing the one or more non-sentiment phrases and the one or more sentiment terms associated with the one or more non-sentiment phrases; and
an analysis unit for periodically determining at least a sentiment trend by performing a correlation analysis on at least two term taxonomies including performing a first correlation and a second correlation, wherein the first correlation is performed on a first stored term taxonomy comprising a non-sentiment phrase and a first sentiment phrase provided by a first group of users and wherein the second correlation is performed on a second stored term taxonomy including said non-sentiment phrase and a second sentiment phrase provided by a second group of users, wherein the first sentiment phrase is different from the second sentiment phrase and the first group of users is different from the second group of users, wherein the analysis unit is further configured to perform a dynamic weighting of trends from different data sources based on at least a frequency that taxonomies results from each data source change.
8. The apparatus of claim 7 , further comprising:
an alert unit for generating an alert responsive of one of: a trend, crossing of a trend threshold, a change in a trend, and a change in frequency occurrence of term taxonomies.
9. The apparatus of claim 7 , wherein the data source is at least one of: a social network, a blog, a web page, and a news feed.
10. The apparatus of claim 7 , wherein the analysis unit further provides a recommendation to increase or decrease advertisement expenditure.
11. The apparatus of claim 10 , wherein the advertisement expenditure is increased or decreased based on the correlation analysis between the at least two term taxonomies.
12. A method for generating a trend report for a non-sentiment phrase, comprising:
generating a plurality of term taxonomies by associating between non-sentiment phrases and sentiment phrases, wherein each of the sentiment phrases includes one or more words describing a sentiment, the sentiment being any one of a positive sentiment, a neutral sentiment, and a negative sentiment, each of the sentiment phrases being associated with a score indicating whether the sentiment phrase describes the sentiment;
storing the plurality of term taxonomies in a non-transitory computer database;
performing periodically at least a statistical analysis respective of the plurality of term taxonomies in the database by performing a correlation analysis including performing a first correlation and a second correlation, wherein the first correlation is performed on a first stored term taxonomy comprising a non-sentiment phrase and a first sentiment phrase provided by a first group of users and wherein the second correlation is performed on a second stored term taxonomy including said non-sentiment phrase and a second sentiment phrase provided by a second group of users, wherein the first sentiment phrase is different from the second sentiment phrase and the first group of users is different from the second group of users;
receiving a request for a report with respect of at least a non-sentiment phrase in the database;
dynamically weighting of trends from one or more data sources based on at least a frequency that taxonomies results from each data source change;
generating a trend report based at least on the at least statistical analysis; and
providing the report to the requestor of the report.
13. The method of claim 12 , wherein generating the plurality of term taxonomies comprises:
crawling the one or more data sources by an agent operative on a computing device to collect textual content from the one or more data sources;
performing phrase extraction using the textual content and generating phrases;
identifying sentiment phrases and non-sentiment phrases from the phrases; and
associating a sentiment phrase with at least a non-sentiment phrase, wherein an association between a non-sentiment phrase and at least one corresponding sentiment phrase is a term taxonomy.
14. The method of claim 13 , wherein identifying the sentiment phrases and non-sentiment phrases comprises:
comparing the phrases to sentiment phrases and non-sentiment phrases stored in a phrases database;
determining that a phrase is a sentiment phrase if a match is found between the phrase and at least one sentiment phrase in the phrase database; and
determining a phrase is a non-sentiment phrase when a match is found between the phrase and at least one non-sentiment phrase in the phrase database.
15. The method of claim 13 , wherein each of the one or more data sources is at least one of: a social network, a blog, a web page, and a news feed.
16. The method of claim 12 , wherein the score is one of: negative, neutral, positive.
17. The method of claim 16 , wherein the negative score is a range of negative values and wherein the positive score is a range of positive values.
18. The method of claim 17 , further comprising:
generating a weighted sentiment score corresponding to a plurality of sentiment phrases collected for a respective non-sentiment phrase.
19. The method of claim 12 , further comprising:
generating an alert responsive of one of: a trend detected from the trend report, crossing of a threshold determined from the trend report, a change in a trend, and a change in a frequency occurrence of term taxonomies.
20. The method of claim 12 , further comprising:
providing a recommendation to increase or decrease advertisement expenditure.
21. The method of claim 20 , wherein the advertisement expenditure is increased or decreased based on the correlation analysis between the at least two term taxonomies to determine their correlation respective of their respective sentiment phrases.
22. A computer software product containing a plurality of instructions embedded in a non-transitory computer readable medium that when executed by a computing device causing to execute the method of claim 12 .