US 12,141,186 B1
Text embedding-based search taxonomy generation and intelligent refinement
Ayush Tomar, Morgan Hill, CA (US); Zengpan Fan, Venice, FL (US); Ketki Savle, Atlanta, GA (US); Fazle Shahnawaz Muhibul Karim, Chicago, IL (US); Ramin Anushiravani, San Carlos, CA (US); and Yizhao Ni, Centennial, CO (US)
Assigned to Optum, Inc., Minnetonka, MN (US)
Filed by Optum, Inc., Minnetonka, MN (US)
Filed on Oct. 11, 2023, as Appl. No. 18/484,943.
Int. Cl. G06F 16/35 (2019.01); G06F 16/383 (2019.01)
CPC G06F 16/355 (2019.01) [G06F 16/383 (2019.01)] 18 Claims
OG exemplary drawing
 
1. A computer-implemented method, the computer-implemented method comprising:
generating, by one or more processors, a plurality of interaction embeddings for a plurality of textual descriptions corresponding to a plurality of interaction codes identified within an interaction dataset;
generating, by the one or more processors, a plurality of taxonomy embeddings for a plurality of taxonomy categories identified within a taxonomy dataset;
generating, by the one or more processors, a plurality of similarity scores for a plurality of description-category pairs based on a comparison between the plurality of interaction embeddings and the plurality of taxonomy embeddings;
generating, by the one or more processors, an enhanced taxonomy by augmenting one or more of the plurality of taxonomy categories with one or more of the plurality of textual descriptions based on the plurality of similarity scores;
initiating, by the one or more processors and using the enhanced taxonomy, performance of a query resolution operation;
generating, by the one or more processors, an evaluation metric for the enhanced taxonomy, wherein the evaluation metric is based on a comparison between (i) a plurality of historical query resolutions that comprises a query resolution associated with the query resolution operation and (ii) interactive user data corresponding to the plurality of historical query resolutions; and
updating, by the one or more processors, one or more of a threshold similarity score or a threshold redundancy value based the evaluation metric.