Assessing complexity of dialogs to streamline handling of service requests
A dialogue complexity assessment method, system, and computer program product for introducing the notion of dialogue complexity to understand and compare dialogues in a repository, calculating the dialogue complexity, use the dialogue complexity to understand customer interactions in a variety of domains using public and proprietary data, and demonstrate the dialogue complexity usage to improve a service management operation.
1. A computer-implemented dialogue complexity assessment method, the method comprising:
calculating a complexity utilizing domain-dependent terms and domain-independent terms of a dialogue in stages by:
computing a complexity of the dialogue at an utterance level;
computing a complexity of the dialogue at a turn level; and
computing a complexity of the dialogue based on the complexity of the constituent turns and utterances,
wherein the complexity calculation distinguishes between any of a domain-dependent term, a common language term, and a stop word in the dialogue.
2. The computer-implemented method of claim 1 , further comprising determining a reason for the complexity of the dialogue, and
wherein the complexity is computed by computing the complexity of the dialogue at the dialogue level based on the utterance level complexity and the turn level complexity.
3. The computer-implemented method of claim 2 , wherein the determining further includes determining rules to explain a reason for dialogue complexity, the rules including a language of the dialogue, a domain of a conversation of the dialogue, and an understandability of a content measured by their uniqueness to the domain.
4. The computer-implemented method of claim 1 , further comprising managing a service handling based on the calculated complexity of the dialogue.
5. The computer-implemented method of claim 4 , wherein the managing determines how a service request is handled at each turn based on the calculated complexity of the dialogue.
6. The computer-implemented method of claim 1 , further comprising ranking a set of service handlers based on the calculated complexity of the dialogue.
7. The computer-implemented method of claim 1 , wherein the complexity uses an N-gram structure.
8. The computer-implemented method of claim 1 , wherein a customer rating of an interaction is weighted with the calculated complexity of the dialogue and a duration of the dialogue, and averaged over a whole duration that an agent is to be evaluated during the dialogue to determine an agent score for a service handler.
9. The computer-implemented method of claim 6 , wherein the service handlers are managed by at least one of:
ranking a set of the service handlers;
assigning an agent to handle service requests;
ranking dialogues in a corpus based on the calculated complexity of the dialogue; and
improving an acquisition of a ground truth for a dialogue system.
10. The computer-implemented method of claim 1 , embodied in a cloud-computing environment.
11. The computer-implemented method of claim 1 , wherein the utterance level, the turn level, and the complexity of the dialogue are calculated and directly used as a dialogue corpus to automatically learn the domain-dependent terms and the domain-independent terms and uses them along with language terms and keywords for the complexity calculation.
12. A computer program product for dialogue complexity assessment, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform:
calculating a complexity utilizing domain-dependent terms and domain-independent terms of a dialogue in stages by:
computing a complexity of the dialogue at an utterance level;
computing a complexity of the dialogue at a turn level; and
computing a complexity of the dialogue based on the complexity of the constituent turns and utterances,
wherein the complexity calculation distinguishes between any of a domain-dependent term, a common language term, and a stop word in the dialogue.
13. The computer program product of claim 12 , further comprising determining a reason for the complexity of the dialogue, and
wherein the complexity is computed by computing the complexity of the dialogue at the dialogue level based on the utterance level complexity and the turn level complexity.
14. The computer program product of claim 13 , wherein the determining further includes determining rules to explain the reason for dialogue complexity, the rules including a language of the dialogue, a domain of a conversation of the dialogue, and an understandability of a content measured by their uniqueness to domain.
15. The computer program product of claim 12 , further comprising managing a service handling based on the calculated complexity of the dialogue.
16. The computer program product of claim 15 , wherein the managing determines how a service request is handled at each turn based on the calculated complexity of the dialogue.
17. The computer program product of claim 12 , further comprising ranking a set of service handlers based on the calculated complexity of the dialogue.
18. The computer program product of claim 12 , wherein the complexity uses an N-gram structure.
19. A dialogue complexity assessment system, said system comprising:
a processor; and
a memory, the memory storing instructions to cause the processor to perform:
calculating a complexity utilizing domain-dependent terms and domain-independent terms of a dialogue in stages by:
computing a complexity of the dialogue at an utterance level;
computing a complexity of the dialogue at a turn level; and
computing a complexity of the dialogue based on the complexity of the constituent turns and utterances,
wherein the complexity calculation distinguishes between any of a domain-dependent term, a common language term, and a stop word in the dialogue.
20. The system of claim 19 , embodied in a cloud-computing environment.