| CPC G06N 5/02 (2013.01) [G06F 16/2291 (2019.01); G06Q 30/0282 (2013.01); G06N 20/00 (2019.01)] | 14 Claims |
|
1. A computer-implemented method, comprising:
receiving, by a computing device, a respective inquiry associated with articles of interest from each of a plurality of user devices;
performing, by a processor, predictive modeling configured to generate training data points based on content of the respective inquiry from each of the plurality of user devices gauging a prospective buyer's level of interest, for the respective inquiry from each of the plurality of user devices, the performing comprising:
receiving:
a first result from a first trained model that indicates an article of interest based on a content element of the respective inquiry that describes content associated with the articles of interest,
a second result from a second trained model that indicates a sentiment for the article of interest based on a sentiment element of the respective inquiry,
a third result from a valuation divergence model that indicates a divergence from a value proposal for the article of interest from a reference value associated with the article of interest based on a value element of the respective inquiry, and
generating a respective composite interest score that indicates a level of interest for the article of interest based on the first result, the second result, and the third result;
providing the respective inquiries from a first set of user devices of the plurality of user devices based on the respective composite interest scores satisfying a scoring threshold to a user to respond to the respective inquiries from the first set of user devices;
providing an indication of another article of interest to a second set of user devices of the plurality of user devices based on the respective composite interest scores being less than the scoring threshold;
generating, by the processor, an automated response to the indication;
determining, by the processor, which of the respective inquiries resulted in a respective sale;
identifying, by the processor, one or more queries associated with each respective inquiry resulting in the respective sale;
identifying, by the processor, one or more content components, from the one or more queries, determined to have been a factor in the respective sale; and
retraining, by the processor, the second trained model based on the one or more content components determined to have been a factor in the respective sale.
|