Enhanced conversational bots processing

1. A computer-implemented method for enhancing user communication with a chatbot, comprising:

receiving a user communication at a chatbot that simulates human conversation and dialogue;

identifying one or more intents related to the user communication;

extracting a plurality of attributes associated with the user communication;

identifying, when an ambiguity in the user communication exists, a plurality of relationships between the extracted attributes using a co-occurrence graph, the co-occurrence graph being a crowd-sourced set of relationships between synonymous terms obtained from a set of training conversations;

determining, in an event that a single intent cannot be ascertained from the co-occurrence graph, an exact relationship between the extracted attributes using an ontology graph that is built based on the co-occurrence graph and continuously updated in response to ongoing evolution of the co-occurrence graph and that contains a set of rules that explicitly encode relationships between the terms; and

generating, based on the plurality of relationships, a disambiguating question that is automatically created from the crowd-sourced set of relationships in the co-occurrence graph and communicated to the user by the chatbot to clarify a user intent of the user communication.

2. The computer-implemented method of claim 1 , further comprising identifying an ambiguity exists in the user communication when more than one viable intent is found.

3. The computer-implemented method of claim 2 , wherein a viable intent includes an intent having a confidence score exceeding a predefined threshold value.

4. The computer-implemented method of claim 1 , further comprising deriving, based on a user answer to the disambiguating question, an actual user intent.

5. The computer-implemented method of claim 4 , further comprising generating, when the actual intent is not derived, another disambiguation question to clarify the user intent.

6. The computer-implemented method of claim 1 , further comprising:

combining the extracted attributes with new input from evolution of an underlying text corpus and from extracted attributes derived from other new user communications received by the chatbot from other users to update the set of relationships in the co-occurrence graph automatically;

building an updated co-occurrence graph by capturing pairs of co-occurring entities in the new entities along with a probability of occurrence;

using a language model to determine whether the pair of co-occurring entities make linguistic sense and to determine weights on edges of the updated co-occurrence graph;

automatically generating a relationship hypothesis based on a new co-occurrence pattern in the updated set of relationships in the updated co-occurrence graph;

building, based on the relationship hypothesis, an algorithm that extracts new synonymy relationships of the updated set of relationships from the updated co-occurrence graph; and

adding the new synonymy relationships extracted from the updated co-occurrence graph to the ontology graph.

7. The computer-implemented method of claim 1 , wherein the user communication is a voice input in a natural language format.

8. A computer program product for enhancing user communication with a chatbot, the computer program product comprising a computer readable storage media, and program instructions stored on the computer readable storage media, that cause at least one computer device to:

receive a user communication at a chatbot that simulates human conversation and dialogue;

identify one or more intents related to the user communication;

extract a plurality of attributes associated with the user communication;

identify, when an ambiguity in the user communication exists, a plurality of relationships between the extracted attributes using a co-occurrence, the co-occurrence graph being a crowd-sourced set of relationships between synonymous terms obtained from a set of training conversations;

determine, in an event that a single intent cannot be ascertained from the co-occurrence graph, an exact relationship between the extracted attributes using an ontology graph that is built based on the co-occurrence graph and continuously updated in response to ongoing evolution of the co-occurrence graph and that contains a set of rules that explicitly encode relationships between the terms; and

generate, based on the plurality of relationships, a disambiguating question that is automatically created from the crowd-sourced set of relationships in the co-occurrence graph and communicated to the user by the chatbot to clarify a user intent of the user communication.

9. The computer program product of claim 8 , the instructions further causing the at least one computer device to identify an ambiguity exists in the user communication when more than one viable intent is found.

10. The computer program product of claim 9 , wherein a viable intent includes an intent having a confidence score exceeding a predefined threshold value.

11. The computer program product of claim 8 , the instructions further causing the at least one computer device to derive, based on a user answer to the disambiguating question, an actual user intent.

12. The computer program product of claim 11 , the instructions further causing the at least one computer device to present the disambiguation question to a user associated with the user communication.

13. The computer program product of claim 8 , the instructions further causing the at least one computer device to:

combine the extracted attributes with new input from evolution of an underlying text corpus and from extracted attributes derived from other new user communications received by the chatbot from other users to update the set of relationships in the co-occurrence graph automatically;

build an updated co-occurrence graph by capturing pairs of co-occurring entities in the new entities along with a probability of occurrence;

use a language model to determine whether the pair of co-occurring entities make linguistic sense and to determine weights on edges of the updated co-occurrence graph;

automatically generate a relationship hypothesis based on a new co-occurrence pattern in the updated set of relationships in the updated co-occurrence graph;

build, based on the relationship hypothesis, an algorithm that extracts new synonymy relationships of the updated set of relationships from the updated co-occurrence graph; and

add the new synonymy relationships extracted from the updated co-occurrence graph to the ontology graph.

14. The computer program product of claim 11 , the instructions further causing the at least one computer device to generate, based on the actual user intent, a response to the user communication.

15. A computer system for enhancing user communication with a chatbot, comprising:

a memory medium comprising instructions;

a bus coupled to the memory medium; and

a processor coupled to the bus that when executing the instructions causes the system to:

receive a user communication at a chatbot that simulates human conversation and dialogue;

identify one or more intents related to the user communication;

extract a plurality of attributes associated with the user communication;

identify, when an ambiguity in the user communication exists, a plurality of relationships between the extracted attributes using a co-occurrence graph, the co-occurrence graph being a crowd-sourced set of relationships between synonymous terms obtained from a set of training conversations;

determine, in an event that a single intent cannot be ascertained from the co-occurrence graph, an exact relationship between the extracted attributes using an ontology graph that is built based on the co-occurrence graph and continuously updated in response to ongoing evolution of the co-occurrence graph and that contains a set of rules that explicitly encode relationships between the terms; and

generate, based on the plurality of relationships, a disambiguating question that is automatically created from the crowd-sourced set of relationships in the co-occurrence graph and communicated to the user by the chatbot to clarify a user intent of the user communication.

16. The computer system of claim 15 , the instructions further causing the system to identify an ambiguity exists in the user communication when more than one viable intent is found.

17. The computer system of claim 16 , wherein a viable intent includes an intent having a confidence score exceeding a predefined threshold value.

18. The computer system of claim 15 , the instructions further causing the system to derive, based on a user answer to the disambiguating question, an actual user intent.

19. The computer system of claim 15 , the instructions further causing the system to generate, when the actual intent is not derived, another disambiguation question to clarify the user intent.

20. The computer system of claim 15 , the instructions further causing the system to:

combine the extracted attributes with new input from evolution of an underlying text corpus and from extracted attributes derived from other new user communications received by the chatbot from other users to update the set of relationships in the co-occurrence graph automatically;

build an updated co-occurrence graph by capturing pairs of co-occurring entities in the new entities along with a probability of occurrence;

use a language model to determine whether the pair of co-occurring entities make linguistic sense and to determine weights on edges of the updated co-occurrence graph;

automatically generate a relationship hypothesis based on a new co-occurrence pattern in the updated set of relationships in the updated co-occurrence graph;

build, based on the relationship hypothesis, an algorithm that extracts new synonymy relationships of the updated set of relationships from the updated co-occurrence graph; and

add the new synonymy relationships extracted from the updated co-occurrence graph to the ontology graph.