Auto-solution help desk advisor
A system includes a processor that executes instructions stored in a memory to implement an auto-solution advisor on a server. The auto-solution advisor receives a current help request describing in lay language text a problem with a computer application or computing device, and determines whether the current help request is textually similar to a previous help request for a previous problem. Based on the similarity of the current help request to the previous help request, the auto-solution advisor assigns a known solution for the previous problem as the suggested solution for the current help request.
1. A computer-implemented method for an automated solution advisor comprised in machine learning based text processing system, the method comprising:
receiving, at a server, a current help request from a user interface at a client computer, the current help request describing in text a problem with a computer application or the computing device;
pre-processing, at the server, the current help request by at least filtering the text to remove one or more non-technical words and stemming the text by reducing one or more technical words in the text to a base form;
in response to the pre-processing, determining, by the server and based on a Latent Dirichlet allocation to provide machine learning based discovery of a plurality of predefined topics, a first frequency distribution of the plurality of predefined topics in the preprocessed current help request;
comparing, by the server, the first frequency distribution of the preprocessed current help request to a plurality of previous help requests, wherein the plurality of previous help requests include the plurality of predefined topics and a plurality of solutions associated with the plurality of predefined topics, wherein the comparing further comprises
determining, by the server, the preprocessed current help request is textually similar to a previous help request, wherein determining the textual similarity between the preprocessed current help request and the previous help request uses a Kullback-Leibler distance (KLD) measure between the first frequency distribution of the preprocessed current help request and a second frequency distribution of the previous help request, wherein the previous help request is associated with a known solution for a previous problem, and wherein the second frequency distribution corresponds to the plurality predefined topics in the previous help request;
assigning, by the server and using the Kullback-Leibler distance (KLD) measure between the preprocessed current help request and the previous help request, the known solution for the previous problem as the known solution for the preprocessed current help request to enable presentation of a response to the current help request; and
in response to a presentation of the known solution at the user interface of the client device, receiving, from the user interface, feedback indicative that the known solution was proper; and
using the feedback to update a database containing the plurality of solutions associated with the plurality of predefined topics.
2. The computer-implemented method of claim 1 , wherein the comparing includes determining, for the plurality of previous help requests, a plurality of frequency distributions.
3. The computer-implemented method of claim 1 , wherein the Kullback-Leibler distance (KLD) measure provides a similarity measure using a term-scoring function based on differences between the first frequency distribution and the second frequency distribution.
4. A computer program product for an automated solution advisor comprised in machine learning based text processing system, the computer program product comprising instructions embodied on a non-transitory computer-readable storage medium, the instructions, when executed by at least one computing device, cause operations comprising:
receiving, at a server, a current help request from a user interface at a client computer, the current help request describing in text a problem with a computer application or the computing device;
pre-processing, at the server, the current help request by at least filtering the text to remove one or more non-technical words and stemming the text by reducing one or more technical words in the text to a base form;
in response to the pre-processing, determining, by the server and based on a Latent Dirichlet allocation to provide machine learning based discovery of a plurality of predefined topics, a first frequency distribution of the plurality of predefined topics in the preprocessed current help request;
comparing, by the server, the first frequency distribution of the preprocessed current help request to a plurality of previous help requests, wherein the plurality of previous help requests include the plurality of predefined topics and a plurality of solutions associated with the plurality of predefined topics, wherein the comparing further comprises
determining, by the server, the preprocessed current help request is textually similar to a previous help request, wherein determining the textual similarity between the preprocessed current help request and the previous help request uses a Kullback-Leibler distance (KLD) measure between the first frequency distribution of the preprocessed current help request and a second frequency distribution of the previous help request, wherein the previous help request is associated with a known solution for a previous problem, and wherein the second frequency distribution corresponds to the plurality predefined topics in the previous help request;
assigning, by the server and using the Kullback-Leibler distance (KLD) measure between the preprocessed current help request and the previous help request, the known solution for the previous problem as the known solution for the preprocessed current help request to enable presentation of a response to the current help request; and
in response to a presentation of the known solution at the user interface of the client device, receiving, from the user interface, feedback indicative that the known solution was proper; and
using the feedback to update a database containing the plurality of solutions associated with the plurality of predefined topics.
5. The computer program product of claim 4 , wherein the comparing includes determining, for the plurality of previous help requests, a plurality of frequency distributions.
6. The computer program product of claim 4 , wherein the Kullback-Leibler distance (KLD) measure provides a similarity measure using a term-scoring function based on differences between the first frequency distribution and the second frequency distribution.
7. A system for an automated solution advisor comprised in machine learning based text processing system, the system comprising:
a processor; and
a memory, the processor executing instructions stored in the memory to cause the system to provide operations comprising:
receiving, at a server, a current help request from a user interface at a client computer, the current help request describing in text a problem with a computer application or the computing device;
pre-processing, at the server, the current help request by at least filtering the text to remove one or more non-technical words and stemming the text by reducing one or more technical words in the text to a base form;
in response to the pre-processing, determining, by the server and based on a Latent Dirichlet allocation to provide machine learning based discovery of a plurality of predefined topics, a first frequency distribution of the plurality of predefined topics in the preprocessed current help request;
comparing, by the server, the first frequency distribution of the preprocessed current help request to a plurality of previous help requests, wherein the plurality of previous help requests include the plurality of predefined topics and a plurality of solutions associated with the plurality of predefined topics, wherein the comparing further comprises
determining, by the server, the preprocessed current help request is textually similar to a previous help request, wherein determining the textual similarity between the preprocessed current help request and the previous help request uses a Kullback-Leibler distance (KLD) measure between the first frequency distribution of the preprocessed current help request and a second frequency distribution of the previous help request, wherein the previous help request is associated with a known solution for a previous problem, and wherein the second frequency distribution corresponds to the plurality predefined topics in the previous help request;
assigning, by the server and using the Kullback-Leibler distance (KLD) measure between the preprocessed current help request and the previous help request, the known solution for the previous problem as the known solution for the preprocessed current help request to enable presentation of a response to the current help request; and
in response to a presentation of the known solution at the user interface of the client device, receiving, from the user interface, feedback indicative that the known solution was proper; and
using the feedback to update a database containing the plurality of solutions associated with the plurality of predefined topics.
8. The system of claim 7 , wherein the comparing includes determining, for the plurality of previous help requests, a plurality of frequency distributions.
9. The system of claim 7 , wherein the Kullback-Leibler distance (KLD) measure provides a similarity measure using a term-scoring function based on differences between the first frequency distribution and the second frequency distribution.