Automatic selection of request handler using trained classification model

1. A computing device comprising:

a hardware processor; and

a machine-readable storage medium storing instructions, the instructions executable by the hardware processor to:

in response to a receipt of an electronic request comprising one or more structured data fields and one or more unstructured data fields, search a database to identify a set of previous electronic requests using the one or more structured data fields of the received electronic request, wherein each of the identified set of previous electronic requests satisfies a threshold of similarity with respect to the one or more structured data fields of the received electronic request;

train a probabilistic classification model using at least one structured data field of the identified set of previous electronic requests;

execute the trained probabilistic classification model using the one or more unstructured data fields of the received electronic request; and

automatically select a request handler using an output of the trained probabilistic classification model.

2. The computing device of claim 1 , wherein the one or more unstructured data fields comprises an unrestricted text field.

3. The computing device of claim 1 , wherein the output of the trained probabilistic classification model comprises a plurality of request handlers, and wherein the instructions cause the hardware processor to:

sort the plurality of request handlers into a sorted order; and

select the request handler based on the sorted order of the plurality of request handlers.

4. The computing device of claim 3 , wherein the plurality of request handlers are sorted based on at least one of handler status, average resolution time, and a number of pending request.

5. The computing device of claim 1 , wherein the instructions further cause the hardware processor to re-train the trained probabilistic classification model in response to:

a determination that an available quantity of request handlers exceeds a handler threshold, and

a determination that an available quantity of previous electronic requests does not exceed a request threshold.

6. The computing device of claim 1 , wherein the electronic request comprises an Information Technology (IT) service request, and wherein the request handler is an IT agent.

7. A non-transitory machine-readable storage medium storing instructions that upon execution cause a processor to:

in response to a receipt of an electronic request comprising a structured data field and an unstructured data field, search a database of previous electronic requests to identify a set of previous electronic requests using the structured data field of the received electronic request, wherein each of the identified set of previous electronic requests satisfies a threshold of similarity with respect to the structured data field of the received electronic request;

train a classification model using at least one structured data field of the identified set of previous electronic requests;

execute the trained classification model using the unstructured data field of the received electronic request; and

select a request handler using an output of the trained classification model.

8. The non-transitory machine-readable storage medium of claim 7 , wherein the classification model is a Naïve Bayes model.

9. The non-transitory machine-readable storage medium of claim 7 , wherein the classification model is a probabilistic classification model, and wherein the instructions further cause the processor to:

train the probabilistic classification model in response to:

a determination that an available quantity of request handlers exceeds a handler threshold, and

a determination that an available quantity of previous electronic requests does not exceed a request threshold.

10. The non-transitory machine-readable storage medium of claim 9 , wherein the instructions further cause the processor to:

in response to a determination that the available quantity of request handlers does not exceed the handler threshold:

execute a pre-trained classification model using the unstructured data field of the received electronic request; and

select the request handler using an output of the pre-trained classification model.

11. The non-transitory machine-readable storage medium of claim 7 , wherein the unstructured data field of the received electronic request comprises a title field or a description field.

12. The non-transitory machine-readable storage medium of claim 7 , wherein the structured data field of the received electronic request comprises a location field or an impact field.

13. The non-transitory machine-readable storage medium of claim 7 , wherein the output of the trained classification model comprises a plurality of request handlers, and wherein the instructions cause the processor to:

sort the plurality of request handlers into a sorted order; and

select the request handler based on the sorted order of the plurality of request handlers.

14. A computer implemented method, comprising:

receiving an electronic request comprising one or more structured data fields and one or more unstructured data fields;

in response to a receipt of the electronic request, searching a database of previous electronic requests to identify a set of previous electronic requests using the one or more structured data fields of the received electronic request, wherein each of the identified set of previous electronic requests satisfies a threshold of similarity with respect to the one or more structured data fields of the received electronic request;

training a probabilistic classification model using at least one structured data field of the identified set of previous electronic requests;

executing the trained probabilistic classification model using the one or more unstructured data fields of the received electronic request; and

selecting a request handler using an output of the trained probabilistic classification model.

15. The computer implemented method of claim 14 , further comprising:

determining whether an available quantity of request handlers exceeds a handler threshold;

determining whether an available quantity of previous electronic requests exceeds a request threshold; and

training the probabilistic classification model in response to a determination that:

the available quantity of request handlers exceeds the handler threshold, and

the available quantity of previous electronic requests does not exceed the request threshold.

16. The computer implemented method of claim 15 , further comprising:

in response to a determination that the available quantity of request handlers does not exceed the handler threshold:

executing a pre-trained classification model using the one or more unstructured data fields of the received electronic request; and

selecting the request handler using an output of the pre-trained classification model.

17. The computer implemented method of claim 15 , further comprising:

in response to a determination that the available quantity of previous electronic requests exceeds the request threshold:

executing a recurrent neural network using the one or more unstructured data fields of the received electronic request; and

selecting the request handler using an output of the recurrent neural network.

18. The computer implemented method of claim 14 , wherein the output of the trained probabilistic classification model comprises a plurality of request handlers, the method further comprising:

sorting the plurality of request handlers into a sorted order; and

selecting the request handler based on the sorted order of the plurality of request handlers.

19. The computer implemented method of claim 18 , wherein the sorted order is based on a quantity of pending requests assigned to each of the plurality of request handlers.