Real-time sentiment analysis for synchronous communication

• US 9,690,775 B2
• Filed: 12/27/2012
• Issued: 06/27/2017
• Est. Priority Date: 12/27/2012
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Fink, Patrick W., McNeil, Kristin E., Parker, Philip E., Werts, David B.
• Primary Examiner(s): Armstrong, Angela A
• Application Number:
  US13/728,898
• Publication Number:
  US 20140188457A1
• Time in Patent Office: 1,643 Days
• CPC Class Codes: G06F 40/30 Semantic analysis

Abstract

A lexical annotator that identifies a chunk of a communication and an associated sentiment is created. In real time, while monitoring a communication from a user, the lexical annotator is used to identify the sentiment for the chunk of the communication, and the sentiment for the chunk of the communication is provided.

Claims

1. A method, comprising:

creating, with a processor of a computer, a lexical annotator that is a rule that identifies a chunk of text and associates a sentiment with that chunk, wherein the chunk is made up of a set of words, and wherein the chunk is received via a user interface;

using a combination of multiple items selected from a group of (1) the lexical annotator, (2) a dictionary entry, and (3) a previously-defined parsing rule annotator to create a new parsing rule annotator; and

in real time, while monitoring a communication in text format from a user,

using the lexical annotator to identify a match of a chunk in the communication to the chunk of the lexical annotator and to identify the sentiment for the chunk of the communication;

storing the sentiment and a position of the match in the communication in a structure;

using the new parsing rule annotator with the sentiment and the position stored in the structure to identify an object of the sentiment;

storing the object of the sentiment in the structure;

providing the sentiment for the chunk of the communication and the object of the sentiment in the structure to a consumer; and

providing a sentiment score for the sentiment, that is selected from sentiment scores by a previously trained machine learning system that is used to automatically score interactions based on at least one of known and learned patterns, wherein scores of phrases are input into the previously trained machine learning system.

2. The method of claim 1, further comprising:

generating a new sentiment for multiple chunks of the communication.

3. The method of claim 1, wherein a portion of the communication is selected for sentiment analysis based on a time period.

4. The method of claim 1, wherein a portion of the communication is selected for sentiment analysis based on a logical grouping of word clusters.

5. The method of claim 1, wherein the communication occurs via at least one of a text message, chat, and telephone call.

6. The method of claim 1, further comprising:

using an Unstructured Information Management Architecture (UIMA) pipeline to identify the sentiment of the chunk and additional information.

7. The method of claim 1, wherein a Software as a Service (SaaS) is provided to perform method operations.

8. A computer program product, the computer program product comprising:

a non-transitory computer readable storage medium having computer readable program code embodied therewith, the computer readable program code, executable by at least one processor of a computer, is configured to perform;

creating a lexical annotator that is a rule that identifies a chunk of text and associates a sentiment with that chunk, wherein the chunk is made up of a set of words, and wherein the chunk is received via a user interface;

using a combination of multiple items selected from a group of (1) the lexical annotator, (2) a dictionary entry, and (3) a previously-defined parsing rule annotator to create a new parsing rule annotator; and

in real time, while monitoring a communication in text format from a user,

using the lexical annotator to identify a match of a chunk in the communication to the chunk of the lexical annotator and to identify the sentiment for the chunk of the communication;

storing the sentiment and a position of the match in the communication in a structure;

using the new parsing rule annotator with the sentiment and the position stored in the structure to identify an object of the sentiment;

storing the object of the sentiment in the structure;

providing the sentiment for the chunk of the communication and the object of the sentiment in the structure to a consumer; and

providing a sentiment score for the sentiment, that is selected from sentiment scores by a previously trained machine learning system that is used to automatically score interactions based on at least one of known and learned patterns, wherein scores of phrases are input into the previously trained machine learning system.

9. The computer program product of claim 8, wherein the computer readable program code, executable by the at least one processor of the computer, is configured to perform:

generating a new sentiment for multiple chunks of the communication.

10. The computer program product of claim 8, wherein a portion of the communication is selected for sentiment analysis based on a time period.

11. The computer program product of claim 8, wherein a portion of the communication is selected for sentiment analysis based on a logical grouping of word clusters.

12. The computer program product of claim 8, wherein the communication occurs via at least one of a text message, chat, and telephone call.

13. The computer program product of claim 8, wherein the computer readable program code, executable by the at least one processor of the computer, is configured to perform:

using an Unstructured Information Management Architecture (UIMA) pipeline to identify the sentiment of the chunk and additional information.

14. The computer program product of claim 8, wherein a Software as a Service (SaaS) is configured to perform computer program product operations.

15. A computer system, comprising:

a processor; and

a storage device connected to the processor, wherein the storage device has stored thereon a program, wherein the processor is configured to execute instructions of the program to perform operations, and wherein the operations comprise;

creating a lexical annotator that is a rule that identifies a chunk of text and associates a sentiment with that chunk, wherein the chunk is made up of a set of words, and wherein the chunk is received via a user interface;

using a combination of multiple items selected from a group of (1) the lexical annotator, (2) a dictionary entry, and (3) a previously-defined parsing rule annotator to create a new parsing rule annotator; and

in real time, while monitoring a communication in text format from a user,

using the lexical annotator to identify a match of a chunk in the communication to the chunk of the lexical annotator and to identify the sentiment for the chunk of the communication;

storing the sentiment and a position of the match in the communication in a structure;

using the new parsing rule annotator with the sentiment and the position stored in the structure to identify an object of the sentiment;

storing the object of the sentiment in the structure;

providing the sentiment for the chunk of the communication and the object of the sentiment in the structure to a consumer; and

providing a sentiment score for the sentiment, that is selected from sentiment scores by a previously trained machine learning system that is used to automatically score interactions based on at least one of known and learned patterns, wherein scores of phrases are input into the previously trained machine learning system.

16. The computer system of claim 15, wherein the operations further comprise:

generating a new sentiment for multiple chunks of the communication.

17. The computer system of claim 15, wherein a portion of the communication is selected for sentiment analysis based on a time period.

18. The computer system of claim 15, wherein a portion of the communication is selected for sentiment analysis based on a logical grouping of word clusters.

19. The computer system of claim 15, wherein the communication occurs via at least one of a text message, chat, and telephone call.

20. The computer system of claim 15, wherein the operations further comprising:

using an Unstructured Information Management Architecture (UIMA) pipeline to identify the sentiment of the chunk and additional information.

21. The computer system of claim 15, wherein a Software as a Service (SaaS) is provided to perform system operations.

22. A computer system for processing a data management request, comprising:

at least one processor; and

hardware logic coupled to the processor, wherein the hardware logic performs operations that comprise;

creating a lexical annotator that is a rule that identifies a chunk of text and associates a sentiment with that chunk, wherein the chunk is made up of a set of words, and wherein the chunk is received via a user interface;

using a combination of multiple items selected from a group of (1) the lexical annotator, (2) a dictionary entry, and (3) a previously-defined parsing rule annotator to create a new parsing rule annotator; and

in real time, while monitoring one or more communications in text format from a user,

analyzing a chunk of the communication to determine a sentiment of the user using a natural language processing framework by;

using the lexical annotator to identify a match of a chunk in the communication to the chunk of the lexical annotator and to identify the sentiment for the chunk of the communication;

storing the sentiment and a position of the match in the communication in a structure;

using the new parsing rule annotator with the sentiment and the position stored in the structure to identify an object of the sentiment;

storing the object of the sentiment in the structure;

providing the sentiment for the chunk of the communication and the object of the sentiment in the structure to a consumer; and

providing a sentiment score for the sentiment, that is selected from sentiment scores by a previously trained machine learning system that is used to automatically score interactions based on at least one of known and learned patterns, wherein scores of phrases are input into the previously trained machine learning system.