Dynamic business rule creation using scored sentiments

1. A method comprising:

sorting, by at least one programmable processor executing a corpus builder, data from a data stream into a plurality of sorted data, the at least one programmable processor being a part of a first computer;

obtaining, by an opinion mining platform executed on at least one programmable processor, a plurality of selected content from the plurality of sorted data based on an identification of a plurality of keywords present in the plurality of sorted data, the plurality of keywords signifying a plurality of sentiments and embedded in the plurality of sorted data;

generating, by the opinion mining platform, selected content, the selected content comprising an amount of data from the data stream associated with and surrounding the plurality of keywords, the amount of data selected sufficient to provide context for further analysis when determining a sentiment that is associated with a keyword of the plurality of keywords, the selected content further comprising metadata associated with the keyword;

cleaning, by the opinion mining platform, the selected content, the cleaning comprising removing data that is not used in any further process;

generating, by a sentiment extraction program executed on the at least one programmable processor, a plurality of sentiment metrics based on an analysis of the plurality of selected content, the analysis being implemented using a hybrid natural language processing technique comprising a machine learning technique and a lexicon-based technique, the hybrid technique improving an accuracy of the analysis, the lexicon-based technique based on a collection of known and precompiled sentiment terms, the plurality of sentiment metrics comprising a score indicating an association of a word of the selected content with a sentiment;

determining, by the sentiment extraction program and based on the plurality of sentiment metrics, the sentiment associated with the selected content; and

performing, by a rules extractor program executed on the at least one programmable processor and in response to the plurality of sentiment metrics satisfying at least one condition of a predetermined rule, an action based on the predetermined rule.

2. The method of claim 1 , wherein the generating the plurality of sentiment metrics further comprises:

generating the selected content by cleaning input from the sorted data;

selecting features from the selected content;

identifying a sentiment of the plurality of sentiments, the sentiment describing the features of the clean data;

scoring and/or classifying the sentiment to determine the sentiment metrics; and

transmitting the sentiment metrics to the rules extractor program.

3. The method of claim 1 wherein, based on the rules extractor program determining that the plurality of conditions are not satisfied, the opinion mining platform continues to obtain the plurality of selected content from the sorted data based on the identification of the plurality of keywords present in the plurality of sorted data.

4. The method of claim 1 wherein, based on the rules extractor program determining that the at least one condition is satisfied, the opinion mining platform continues to obtain the plurality of selected content from the sorted data based on the identification of the plurality of keywords present in the plurality of sorted data.

5. The method of claim 1 , wherein the sentiment comprises a subject entity about whom/which the sentiment is expressed, a feature of the subject entity, and a subjective opinion about the feature of the subject.

6. The method of claim 1 , wherein the plurality of sentiment metrics are numerical values.

7. The method of claim 1 , wherein the plurality of sentiment metrics are descriptive values.

8. The method of claim 1 , wherein the business rules engine is fully automated and does not require human intervention for execution of the action.

9. The method of claim 1 , wherein the business rules engine is partially automated and requires human intervention for execution of the action.

10. The method of claim 1 , wherein the selected content comprises text, video, audio, webpage information, and/or metadata.

11. The method of claim 1 , wherein the plurality of actions is performed by the business rules engine.

12. A computer program product comprising a non-transitory machine-readable medium storing instructions that, when executed by at least one programmable processor, cause the at least one programmable processor of a first computer to perform operations comprising:

sorting, by a corpus builder, data from a data stream into a plurality of sorted data;

obtaining, by an opinion mining platform, a plurality of selected content from the plurality of sorted data based on an identification of a plurality of keywords present in the plurality of sorted data, the plurality of keywords signifying a plurality of sentiments and embedded in the plurality of sorted data;

generating, by the opinion mining platform, selected content, the selected content comprising an amount of data from the data stream associated with and surrounding the plurality of keywords, the amount of data selected sufficient to provide context for further analysis when determining a sentiment that is associated with a keyword of the plurality of keywords, the selected content further comprising metadata associated with the keyword;

cleaning, by the opinion mining platform, the selected content, the cleaning comprising removing data that is not used in any further process;

generating, by a sentiment extraction program, a plurality of sentiment metrics based on an analysis of the plurality of selected content, the analysis being implemented using a hybrid natural language processing technique comprising a machine learning technique and a lexicon-based technique, the hybrid technique improving an accuracy of the analysis, the lexicon-based technique based on a collection of known and precompiled sentiment terms, the plurality of sentiment metrics comprising a score indicating an association of a word of the selected content with a sentiment;

determining, by the sentiment extraction program and based on the plurality of sentiment metrics, the sentiment associated with the selected content; and

performing, by a rules extractor program executed on the at least one programmable processor and in response to the plurality of sentiment metrics satisfying at least one condition of a predetermined rule, an action based on the predetermined rule.

13. The computer program product of claim 12 , wherein the generating further comprises:

generating the selected content by cleaning input from the sorted data;

selecting features from the selected content;

identifying a sentiment of the plurality of sentiments, the sentiment describing the features of the clean data;

scoring and/or classifying the sentiment to determine the sentiment metric; and

transmitting the sentiment metric to the rules extractor program.

14. The computer program product of claim 12 , wherein the plurality of actions is performed by the business rules engine.

15. The computer program product of claim 12 , wherein the plurality of actions is not performed by the business rules engine.

16. The computer program product of claim 12 , wherein irrespective of the determining, the opinion mining platform continues to obtain the plurality of selected content from the sorted data based on the identification of the plurality of keywords present in the plurality of sorted data.

17. A system comprising:

at least one programmable processor of a first computer; and

memory storing instructions, which when executed by the at least one programmable processor, result in operations comprising:

sorting, by at least one programmable processor executing a corpus builder, data from a data stream into a plurality of sorted data;

obtaining, by an opinion mining platform executed on at least one programmable processor, a plurality of selected content from the plurality of sorted data based on an identification of a plurality of keywords present in the plurality of sorted data, the plurality of keywords signifying a plurality of sentiments embedded in the plurality of sorted data, the plurality of selected content comprising an amount of data from the data stream associated with and surrounding the plurality of keywords, the amount of data selected sufficient to provide context for further analysis when determining a sentiment that is associated with a keyword of the plurality of keywords, the selected content further comprising metadata associated with the keyword;

cleaning, by the opinion mining platform, the selected content, the cleaning comprising removing data that is not used in any further process;

generating, by a sentiment extraction program executed on the at least one programmable processor, a plurality of sentiment metrics based on an analysis of the plurality of selected content, the analysis being implemented using a hybrid natural language processing technique comprising a machine learning technique and a lexicon-based technique, the hybrid technique improving an accuracy of the analysis, the lexicon-based technique based on a collection of known and precompiled sentiment terms, the plurality of sentiment metrics comprising a score indicating an association of a word of the selected content with a sentiment;

determining, by the sentiment extraction program and based on the plurality of sentiment metrics, the sentiment associated with the selected content; and

performing, by a rules extractor program executed on the at least one programmable processor and in response to the plurality of sentiment metrics satisfying at least one condition of a predetermined rule, an action based on the predetermined rule.

18. The system of claim 17 , further comprising generating the selected content by cleaning input from the sorted data.

19. The system of claim 17 , further comprising selecting features from the selected content.

20. The system of claim 17 , further comprising identifying a sentiment of the plurality of sentiments, the sentiment describing the features of the clean data.

21. The system of claim 17 , further comprising scoring and/or classifying the sentiment to determine the sentiment metrics.

22. The system of claim 17 , further comprising transmitting the sentiment metrics to the rules extractor program.