System and method of rule creation based on frequency of question
A method is provided for prioritizing rule creation for computer-assisted customer care. When a question from a user of a device is received by a customer care analytics engine for which no rule is automatically fired, the question and a related device profile of the device are added to an unfired questions list. The analytics engine parses the question to match terms in other questions in the list. The question is also added to an appropriate category in the list based on the device profile. A prioritization algorithm is used to rank the question among other questions within the list or the category, ranking the question more highly according to the frequency of those terms in the category or the list. According to its rank-wise order, the question is directed for creation of a rule to permit automatic handling of questions having the same or similar terms in the future.
1. A method of prioritizing rule creation for computer-assisted customer care to a user of a device, comprising:
on receipt by a customer care analytics engine of a question from the user, together with a device profile of the device, the question being one for which no rule was automatically fired by a rules engine, adding the question to an unfired questions list;
the customer care analytics engine parsing the question to match terms in other questions in the list;
adding the question to a category in the list based on the device profile;
applying a prioritization algorithm to rank the question among other questions within the list or the category, the algorithm ranking the question more highly according to the frequency of those terms in the category or the list; and
according to its rank-wise order, directing the question to a rules engine for creation of a rule to permit automatic handling of questions having the same terms in the future, and ignoring questions whose terms are below a frequency threshold.
2. The method of claim 1 , wherein the prioritization algorithm is applied to a snapshot of questions.
3. The method of claim 2 , wherein the snapshot is periodic.
4. The method of claim 2 , wherein the snapshot is based on an event.
5. The method of claim 2 , wherein the snapshot is based on a threshold.
6. The method of claim 1 , wherein the parsing includes natural language processing.
7. The method of claim 1 , wherein the parsing includes latent semantic analysis.
8. The method of claim 1 , wherein the parsing identifies terms related to negative events.
9. The method of claim 1 , wherein the parsing identifies terms related to device functions.
10. The method of claim 1 , wherein the frequency of terms is evaluated using a document term matrix to count how often each term occurs across questions in the category or the list.
11. The method of claim 1 , further comprising referring the question to a customer support representative (CSR) for resolution.
12. The method of claim 11 , wherein the time to resolution for the question is calculated.
13. The method of claim 12 , wherein the prioritization algorithm includes comparing the time to resolution of the question to a mean time to resolution (MTTR) of other questions in the list.
14. The method of claim 13 , wherein the ranking in the prioritization algorithm includes weighting the questions by MTTR.
15. The method of claim 1 , wherein the question is submitted by text or voice.
16. The method of claim 11 , wherein any changes to the device profile in the course of the CSR resolution are tracked.
17. The method of claim 11 , wherein any discussion or text exchanges between the CSR and the user during the course of the resolution are tracked.