System for augmenting conversational system training with reductions

1. A computer-implemented method for augmenting training of a conversational system, comprising:

ingesting a training set to be used in training of the conversational system, the training set including objects for use in the training, wherein the objects include one or more object types, wherein the object types include one or more of an entity category or intent category;

generating proposed reductions for inclusion in an augmented training set, wherein the proposed reductions include one or more of:

reduction candidates generated using properties of objects included in the training set;

reduction candidates obtained from search queries of one or more external resources, wherein the search queries relate to one or more objects included in the training set;

filtering the proposed reductions to generate a reduced set of proposed reductions which comprises:

assigning a confidence level to each of the reduction candidates; and

prioritizing the reduction candidates using the confidence levels; and

augmenting the training set with the reduced set of proposed reductions or keeping only the reduced set of proposed reductions.

2. The computer-implemented method of claim 1 , wherein reduction candidates generated using properties of entities included in the training set comprise:

identifying objects having sequential capitalization of words in a row, wherein

a proposed reduction for the identified object is generated using capitalized characters included in the sequential capitalization of the words, and

disregarding generation of a proposed reduction for the identified object when the sequential capitalization spans a sentence break.

3. The computer-implemented method of claim 1 , wherein filtering the proposed reductions comprises one or more of:

removing proposed reductions having synthetically harvested inappropriate terms;

validating proposed reductions using domain specific corpus; and

validating proposed reductions using non-domain specific corpus;

validating proposed reductions based on usage of the proposed reductions located on external resources.

4. The computer-implemented method of claim 1 , wherein the external resources comprise:

social media streams;

domain specific search engines; and

websites relating to an entity included in the training set.

5. The computer-implemented method of claim 1 , wherein the proposed reductions include:

reductions extracted through heuristic analysis of augmented communications systems.

6. A system comprising:

a processor;

a data bus coupled to the processor; and

a computer-usable medium embodying computer program code, the computer-usable medium being coupled to the data bus, the computer program code comprising instructions executable by the processor and configured for:

ingesting a training set to be used in training of the conversational system, the training set including objects for use in the training, wherein the objects include one or more object types, wherein the object types include one or more of an entity category or intent category;

generating proposed reductions for inclusion in an augmented training set, wherein the proposed reductions include one or more of:

reduction candidates generated using properties of objects included in the training set;

reduction candidates obtained from search queries of one or more external resources, wherein the search queries relate to one or more objects included in the training set;

filtering the proposed reductions to generate a reduced set of proposed reductions which comprises:

assigning a confidence level to each of the reduction candidates; and

prioritizing the reduction candidates using the confidence levels; and

augmenting the training set with the reduced set of proposed reductions or keeping only the reduced set of proposed reductions.

7. The system of claim 6 , wherein reduction candidates generated using properties of entities included in the training set comprise:

identifying objects having sequential capitalization of words in a row, wherein

a proposed reduction for the identified object is generated using capitalized characters included in the sequential capitalization of the words, and

inhibiting generation of a proposed reduction for the identified object when the sequential capitalization spans a sentence break.

8. The system of claim 6 , wherein filtering the proposed reductions comprises one or more of:

removing proposed reductions having synthetically harvested inappropriate terms;

validating proposed reductions using domain specific corpus; and

validating proposed reductions using non-domain specific corpus;

validating proposed reductions based on usage of the proposed reductions located on external resources.

9. The system of claim 6 , wherein the external resources comprise:

social media streams;

domain specific search engines; and

websites relating to an entity included in the training set.

10. The system of claim 6 , wherein the proposed reductions include:

reductions extracted through heuristic analysis of augmented communications systems.

11. The system of claim 6 , wherein filtering the proposed reductions further comprises:

assigning a confidence level to each of the reduction candidates; and

prioritizing the reduction candidates using the confidence levels.

12. A non-transitory, computer-readable storage medium embodying computer program code, the computer program code comprising computer executable instructions configured for:

ingesting a training set to be used in training of the conversational system, the training set including objects for use in the training, wherein the objects include one or more object types, wherein the object types include one or more of an entity category or intent category;

generating proposed reductions for inclusion in an augmented training set, wherein the proposed reductions include one or more of:

reduction candidates generated using properties of objects included in the training set;

reduction candidates obtained from search queries of one or more external resources, wherein the search queries relate to one or more objects included in the training set;

filtering the proposed reductions to generate a reduced set of proposed reductions which comprises:

assigning a confidence level to each of the reduction candidates; and

prioritizing the reduction candidates using the confidence levels; and

augmenting the training set with the reduced set of proposed reductions or keeping only the reduced set of proposed reductions.

13. The non-transitory, computer-readable storage medium of claim 12 , wherein filtering the proposed reductions comprises one or more of:

removing proposed reductions having synthetically harvested inappropriate terms;

validating proposed reductions using domain specific corpus; and

validating proposed reductions using non-domain specific corpus;

validating proposed reductions based on usage of the proposed reductions located on external resources.

14. The non-transitory, computer-readable storage medium of claim 12 , wherein the external resources comprise:

social media streams;

domain specific search engines; and

websites relating to an entity included in the training set.

15. The non-transitory, computer-readable storage medium of claim 12 , wherein the proposed reductions include:

reductions extracted through heuristic analysis of augmented communications systems.

16. The computer-implemented method of claim 1 , wherein filtering the proposed reductions further comprises:

assigning a confidence level to each of the reduction candidates; and

prioritizing the reduction candidates using the confidence levels.

17. The computer-implemented method of claim 1 , wherein filtering the proposed reductions further comprises:

assigning a confidence level to each of the reduction candidates; and

prioritizing the reduction candidates using the confidence levels.