Detection and labeling of conversational actions

Patent metadata

Field	Value
Publication number	US-9817817-B2
Application number	US-201615072848-A
Country	US
Kind code	B2
Filing date	Mar 17, 2016
Priority date	Mar 17, 2016
Publication date	Nov 14, 2017
Grant date	Nov 14, 2017

Abstract

Technical solutions are described for generating structured conversational data. An example method includes receiving an utterance that is part of a conversation and identifying the utterance as part of an adjacency pair. The adjacency pair includes two utterances, each produced by different speakers. The method also includes associating the utterance with a label from a predetermined set of labels based on the identifying of the adjacency pair.

First claim

What is claimed is:

1. A computer implemented method for generating structured conversational data, the
method comprising:
receiving, by a processor, an utterance that is part of a conversation;
identifying, by the processor, the utterance as part of an adjacency pair, wherein the adjacency pair comprises two utterances, each produced by different speakers; and
associating, by the processor, the utterance with a label from a predetermined set of labels based on the identifying of the adjacency pair, and wherein associating the utterance with the label further comprises:
determining if the adjacency pair is a base pair; and in response to the adjacency pair being a base pair, selecting the label corresponding either to a first pair-part of the base pair or a second pair-part of the base pair.

2. The computer implemented method of claim 1 , wherein the predetermined set of labels, which comprises a plurality of labels corresponding to respective positions according to an adjacency pair model.

3. The computer implemented method of claim 1 , wherein the method further comprises, initiating, by the processor, the conversation between a machine and a human user and the adjacency pair includes one utterance by the machine and one utterance from the human user.

4. The computer implemented method of claim 1 , wherein the method further includes, determining a type of the adjacency pair and wherein, the label of the utterance is further based on the type of the adjacency pair.

5. The computer implemented method of claim 1 , wherein associating the utterance with the label further comprises:
determining if the adjacency pair is a pre-expansion pair; and
in response to the adjacency pair being a pre-expansion pair, selecting the label corresponding either to a first pair-part of the pre-expansion pair or a second pair-part of the pre-expansion pair.

6. The computer implemented method of claim 1 , wherein associating the utterance with the label further comprises:
determining if the adjacency pair is an insert expansion pair; in response to the adjacency pair being an insert expansion pair:
determining if the adjacency pair is a second position repair sequence;
in response to the adjacency pair not being a second position repair sequence selecting the label corresponding either to a first pair-part of the insert expansion pair or a second pair-part of the insert expansion pair; and in response to the adjacency pair being a second position repair sequence selecting the label corresponding either to an initiator part of the second position repair pair or a repair part the second position repair pair.

7. The computer implemented method of claim 6 , wherein associating the utterance with the label further comprises:
in response to the adjacency pair not being an insert expansion pair and the adjacency pair not being a repair pair, selecting the label corresponding either to a first pair-part of a post-expansion pair or a second pair-part of the post-expansion pair.

8. The computer implemented method of claim 6 , wherein associating the utterance with the label further comprises:
in response to the adjacency pair not being an insert expansion pair and the adjacency pair being a repair pair determining a type of repair performed by the adjacency pair.

9. A system for generating structured conversational data,
the system comprising:
a memory; and a processor configured to:
receive an utterance that is part of a conversation;
identify the utterance as part of an adjacency pair, wherein the adjacency pair comprises two utterances, each produced by different speakers; and
associate the utterance with a label from a predetermined set of labels based on the identifying of the adjacency pair, wherein associating the utterance with the label further comprises:
determining if the adjacency pair is a base pair; and
in response to the adjacency pair being a base pair, selecting the label corresponding either to a first pair-part of the base pair or a second pair-part of the base pair.

10. The system of claim 9 , wherein the processor is further configured to initiate the conversation between a machine and a human user and the adjacency pair includes one utterance by the machine and one utterance from the human user.

11. The system of claim 9 , wherein the processor is further configured to determine a type of the adjacency pair and wherein, the label of the utterance is further based on the type of the adjacency pair.

12. The system of claim 9 , wherein associating the utterance with the label further comprises:
determining if the adjacency pair is a pre-expansion pair; and
in response to the adjacency pair being a pre-expansion pair, selecting the label corresponding either to a first pair-part of the pre-expansion pair or a second pair-part of the pre-expansion pair.

13. The system of claim 9 , wherein associating the utterance with the label further comprises:
determining if the adjacency pair is an insert expansion pair;
in response to the adjacency pair being an insert expansion pair:
determining if the adjacency pair is a second position repair sequence;
in response to the adjacency pair not being a second position repair sequence selecting the label corresponding either to a first pair-part of the insert expansion pair or a second pair-part of the insert expansion pair; and
in response to the adjacency pair being a second position repair sequence selecting the label corresponding either to an initiator part of the second position repair pair or a repair part the second position repair pair.

14. The system of claim 13 , wherein associating the utterance with the label further comprises:
in response to the adjacency pair not being an insert expansion pair and the adjacency pair not being a repair pair, selecting the label corresponding either to a first pair-part of a post-expansion pair or a second pair-part of the post-expansion pair.

15. A computer program product for generating structured conversational data, the computer program product comprising a computer readable storage medium, the computer readable storage medium comprising computer executable instructions, wherein the computer readable storage medium comprises instructions for:
receiving, by a processor, an utterance that is part of a conversation;
identifying, by the processor, the utterance as part of an adjacency pair, wherein the adjacency pair comprises two utterances, each produced by different speakers; and
associating, by the processor, the utterance with a label from a predetermined set of labels based on the identifying of the adjacency pair, wherein the associating comprises determining a type of the adjacency pair, the label of the utterance is being based on the type of the adjacency pair.

16. The computer program product of claim 15 , wherein the type of the adjacency pair is one of a base pair, a pre-expansion pair, an insert expansion pair, a post-expansion pair, and a repair pair.

17. The computer program product of claim 15 , wherein associating the utterance with the label further comprises:
determining if the adjacency pair is a base pair, and in response to the adjacency pair being a base pair, selecting the label corresponding either to a first pair-part of the base pair or a second pair-part of the base pair;
in response to the adjacency pair not being a base pair, determining if the adjacency pair is a pre-expansion pair, and in response to the adjacency pair being a preexpansion pair, selecting the label corresponding either to a first pair-part of the preexpansion pai

Assignees

• IBM

Inventors

• Hosn Rafah A
• Moore Robert J
• Shmueli-Scheuer Michal

Classifications

• G06F40/117 Primary

  Tagging; Marking up (details of markup languages G06F40/143); Designating a block; Setting of attributes (style sheets, e.g. eXtensible Stylesheet Language Transformation [XSLT], G06F40/154) · CPC title

• G06F40/35 Primary

  Discourse or dialogue representation · CPC title

• G06F17/279 Primary

  Physics · mapped topic

• G06F17/2775

  Physics · mapped topic