| CPC G06F 40/295 (2020.01) [G06F 40/284 (2020.01)] | 20 Claims |
|
1. A system comprising:
one or more memories storing computer-executable instructions; and
one or more hardware processors to execute the computer-executable instructions to:
access text data that includes first text describing an item and second text indicative of a plurality of item characteristics that include at least a first item characteristic and a second item characteristic to be located within the first text; and
provide the text data as a single input to a machine learning model and extract information regarding each of the plurality of item characteristics in a single pass by:
determining a first set of output embeddings based on the first text;
determining a second set of output embeddings based on the second text;
determining a first portion of the second set of output embeddings that is associated with the first item characteristic;
determining, based on the first set of output embeddings and the first portion of the second set of output embeddings, a first set of interaction embeddings;
determining, using the first set of interaction embeddings, a first set of label predictions that indicate at least one first embedding of the first set of interaction embeddings as being associated with the first item characteristic;
determining a second portion of the second set of output embeddings that is associated with the second item characteristic;
determining, based on the first set of output embeddings and the second portion of the second set of output embeddings, a second set of interaction embeddings;
determining, using the second set of interaction embeddings, a second set of label predictions that indicate at least one second embedding of the second set of interaction embeddings as being associated with the second item characteristic; and
generating output that associates the at least one first embedding with the first item characteristic and the at least one second embedding with the second item characteristic.
|