Facilitating detection of conversation threads in a messaging channel

Patent metadata

Field	Value
Publication number	US-11263402-B2
Application number	US-201916404156-A
Country	US
Kind code	B2
Filing date	May 6, 2019
Priority date	May 6, 2019
Publication date	Mar 1, 2022
Grant date	Mar 1, 2022

Abstract

Techniques facilitating detection of conversation threads in unstructured channels are provided. A system can comprise a memory that stores computer executable components and a processor that executes the computer executable components stored in the memory. The computer executable components can comprise an extraction component that employs a model to detect conversation messages based on a defined confidence level and assigns the conversation messages to respective conversation thread categories. The computer executable components also can comprise a model component that trains the model on conversation messages that comprise respective text data, wherein the model is trained to detect the respective text data to the defined confidence level.

First claim

What is claimed is:

1. A system, comprising:
a memory that stores computer executable components; and
a processor that executes the computer executable components stored in the memory, wherein the computer executable components comprise:
a model component that trains, using machine learning, a model on a first set of unstructured conversation messages that comprise respective text data, wherein the model is trained to detect the respective text data to a defined confidence level; and
an extraction component that employs the model to transform a second set of unstructured conversation messages from a plurality of parties communicating in a communication channel into structured conversation threads having respective contexts, wherein the model:
for a first subset of first unstructured conversation messages of the second set of unstructured conversation messages performs a pairwise sentence comparison of the first unstructured conversation messages to assign the first unstructured conversation messages to the structured conversation threads, and
for a second subset of second unstructured conversation messages of the second set of unstructured conversation messages:
performs a context similarity comparison of the second unstructured conversation messages to the respective contexts of the structured conversation threads, and
based on the context similarity comparison, for respective second unstructured conversation messages, at least one of assign the the second unstructured conversation message to one of the structured conversation threads or assign the second unstructured conversation message to a new structured conversation thread, wherein the context similarity comparison does not comprise the pairwise sentence comparison involving the second unstructured conversation messages, and the first subset is different than the second subset.

2. The system of claim 1 , wherein the context similarity comparison generates respective context scores between the second unstructured conversation messages and the respective contexts.

3. The system of claim 1 , wherein the model component trains the model to identify the respective text data of the first set of unstructured conversation messages received over a defined interval.

4. The system of claim 1 , wherein the model component trains the model on the respective text data of the first set of unstructured conversation messages that commenced during a defined interval.

5. The system of claim 4 , wherein the model component ignores prior unstructured conversation messages that commenced prior to the defined interval for training the model.

6. The system of claim 1 , wherein the second set of unstructured conversation messages comprise parallel conversations that occur during an overlapping time period.

7. The system of claim 1 , wherein the pairwise sentence comparison generates a sentence similarity score.

8. The system of claim 1 , wherein the performing the context similarity comparison comprises:
performing the context similarity comparison of the second unstructured conversation messages to the respective contexts of the structured conversation threads, comprising:
in response to the second unstructured conversation message matching, according to a similarity criterion, a context of the respective contexts, assigning the second unstructured conversation message to a first structured conversation thread associated with the context, and
in response to the second unstructured conversation message not matching, according to the similarity criterion, the respective contexts, assigning the second unstructured conversation message to the new structured conversation thread associated with a new context.

9. A computer-implemented method, comprising:
training, by a system operatively coupled to a processor, using machine learning, a model on a first set of unstructured conversation messages that comprise respective text data, wherein the model is trained to detect the respective text data to a defined confidence level;
employing, by the system, the model on a first subset of first unstructured conversation messages of a second set of unstructured conversation messages from a plurality of parties communicating in a communication channel to perform a pairwise sentence comparison of the first unstructured conversation messages to assign the first unstructured conversation messages to structured conversation threads; and
employing, by the system, the model on a second subset of second unstructured conversation messages of the second set of unstructured conversation messages to:
perform a context similarity comparison of the second unstructured conversation messages to respective contexts of the structured conversation threads, and
based on the context similarity comparison, for respective second unstructured conversation messages, at least one of assign the second unstructured conversation message to one of the structured conversation threads or assign the second unstructured conversation message to a new structured conversation thread, wherein the context similarity comparison does not comprise the pairwise sentence comparison involving the second unstructured conversation messages, and the first subset is different than the second subset.

10. The computer-implemented method of claim 9 , wherein the context similarity comparison generates respective context scores between the second unstructured conversation messages and the respective contexts.

11. The computer-implemented method of claim 9 , wherein the training the model further comprises training, by the system using the machine learning, the model to identify the respective text data of the first set of unstructured conversation messages received over a defined interval.

12. The computer-implemented method of claim 9 , wherein the training the model further comprises training, by the system, the model on the respective text data of the first set of unstructured conversation messages that commenced during a defined interval.

13. The computer-implemented method of claim 12 , further comprising:
ignoring, by the system, prior unstructured conversation messages that commenced prior to the defined interval for training the model.

14. The computer-implemented method of claim 9 , wherein the second set of unstructured conversation messages comprise parallel conversations that occur during an overlapping time period.

15. The computer-implemented method of claim 9 , wherein the pairwise sentence comparison generates a sentence similarity score.

16. The computer-implemented method of claim 9 , wherein the performing the context similarity comparison comprises:
performing, by the system using the model, the context similarity comparison of the second unstructured conversation messages to the respective contexts of the structured conversation threads, comprising:
in response to the second unstructured conversation message matching, according to a similarity criterion, a context of the respective contexts, assigning the second unstructured conversation message to a first structured conversation thread associated with the context, and
in response to the second unstructured conversation message not matching, according to the similarity criterion, the respective contexts, assigning the second unstructured conversation message to the new structured conversation thread associated with a new context.

17. A computer program product that facilitates detection of conversation threads in a messaging channel, the computer program product comprising a computer readable storage medium having program

Assignees

• IBM

Inventors

• Tan Ming
• Wang Dakuo
• Yu Mo
• Gan Chuang
• Wang Haoyu
• Chang Shiyu

Classifications

• G06F40/289

  Phrasal analysis, e.g. finite state techniques or chunking · CPC title

• G06F40/30 Primary

  Semantic analysis · CPC title

• G06F40/205

  Parsing · CPC title

• G06F40/35 Primary

  Discourse or dialogue representation · CPC title