Techniques for reinforcement for bots using capability catalogs
Techniques for reinforcement for bots are described. In one embodiment, an apparatus may comprise a bot application interface component operative to receive a plurality of bot capability catalogs for a plurality of bots at a bot-service system; a client communication component operative to receive a plurality of user service prompts from a plurality of user client devices; and an interaction processing component operative to determine the selected bots of the plurality of bots for each of the plurality of user prompts by matching the plurality of user service prompts against the plurality of bot capability catalogs using a bot capability table generated by a natural-language machine-learning component; record a bot interaction history based on user interactions with the selected bots; and update the natural-language machine-learning component based on the bot interaction history. Other embodiments are described and claimed.
1. A computer-implemented method, comprising:
receiving a plurality of bot capability catalogs expressing bot capabilities in natural language;
receiving a plurality of user service prompts in natural language;
selecting one or more bots for each user service prompt by performing a similarity search on a plurality of bot capability vectors representing the bot capability catalogs using, as a search term, a user intent vector representing user intent of the user service prompt, the user intent vector having been translated from the natural-language user service prompt via one or more natural-language-to-vector techniques, wherein the bot capability vectors are included in a bot capability table generated by a natural-language machine-learning component, wherein the bot capability and user intent vectors are numerical vectors, and wherein the similarity search is based at least in part on a mathematical similarity comparison;
providing the selected one or more bots for each user service prompt to a plurality of user client devices;
recording a bot interaction history based on user interactions with the selected one or more bots; and
updating the natural-language machine-learning component based on the bot interaction history.
2. The method of claim 1 , further comprising:
updating the natural-language machine-learning component based on optimizing for bot completion rate.
3. The method of claim 2 , wherein optimizing for bot completion rate is based on user bot dismissal rates.
4. The method of claim 2 , wherein optimizing for bot completion rate is based on maintaining a user bot dismissal rate below a defined bot dismissal threshold.
5. The method of claim 4 , wherein optimizing for bot completion rate is based on user-bot-history signals.
6. The method of claim 1 , further comprising:
updating the natural-language machine-learning component based on increasing bot information sharing.
7. The method of claim 1 , further comprising:
updating the natural-language machine-learning component to modify the mapping between the plurality of bot capability catalogs and the plurality of bot capability vectors.
8. An apparatus, comprising:
at least one processor operable to execute stored instructions that, when executed, cause the at least one processor to:
receive a plurality of bot capability catalogs expressing bot capabilities in natural language;
receive a plurality of user service prompts in natural language;
select one or more bots for each user service prompt based on a similarity search on a plurality of bot capability vectors representing the bot capability catalogs using, as a search term, a user intent vector representing user intent of the user service prompt, the user intent vector having been translated from the natural-language user service prompt via one or more natural-language-to-vector techniques, wherein the bot capability vectors are included in a bot capability table generated by a natural-language machine-learning component, wherein the bot capability and user intent vectors are numerical vectors, and wherein the similarity search is based at least in part on a mathematical similarity comparison;
provide the selected one or more bots for each user service prompt to a plurality of user client devices;
record a bot interaction history based on user interactions with the selected one or more bots; and
update the natural-language machine-learning component based on the bot interaction history.
9. The apparatus of claim 8 ,
wherein the at least one processor is further caused to update the natural-language machine-learning component based on optimizing for bot completion rate.
10. The apparatus of claim 9 , wherein optimizing for bot completion rate is based on user bot dismissal rates.
11. The apparatus of claim 9 , wherein optimizing for bot completion rate is based on maintaining a user bot dismissal rate below a defined bot dismissal threshold.
12. The apparatus of claim 11 , wherein optimizing for bot completion rate is based on user-bot-history signals.
13. The apparatus of claim 8 ,
wherein the at least one processor is further caused to update the natural-language machine-learning component based on increasing bot information sharing.
14. The apparatus of claim 8 , wherein
the at least one processor is further caused to update the natural-language machine-learning component to modify the mapping between the plurality of bot capability catalogs and the plurality of bot capability vectors.
15. At least one computer-readable storage medium comprising instructions that, when executed, cause a system to:
receive a plurality of bot capability catalogs expressing bot capabilities in natural language;
receive a plurality of user service prompts in natural language;
selecting one or more bots for each user service prompt by based on a similarity search on a plurality of bot capability vectors representing the bot capability catalogs using, as a search term, a user intent vector representing user intent of the user service prompt, the user intent vector having been translated from the natural-language user service prompt via one or more natural-language-to-vector techniques, wherein the bot capability vectors are included in a bot capability table generated by a natural-language machine-learning component, wherein the bot capability and user intent vectors are numerical vectors, and wherein the similarity search is based at least in part on a mathematical similarity comparison;
provide the selected one or more bots for each user service prompt to a plurality of user client devices;
record a bot interaction history based on user interactions with the selected one or more bots; and
update the natural-language machine-learning component based on the bot interaction history.
16. The computer-readable storage medium of claim 15 , comprising further instructions that, when executed, cause the system to:
update the natural-language machine-learning component based on optimizing for bot completion rate.
17. The computer-readable storage medium of claim 16 , wherein optimizing for bot completion rate is based on user bot dismissal rates.
18. The computer-readable storage medium of claim 16 , wherein optimizing for bot completion rate is based on maintaining a user bot dismissal rate below a defined bot dismissal threshold, wherein optimizing for bot completion rate is based on user-bot-history signals.
19. The computer-readable storage medium of claim 15 , comprising further instructions that, when executed, cause the system to:
update the natural-language machine-learning component based on increasing bot information sharing.
20. The computer-readable storage medium of claim 15 , comprising further instructions that, when executed, cause the system to:
update the natural-language machine-learning component to modify the mapping between the plurality of bot capability catalogs and the plurality of bot capability vectors.