CONTINUOUS LEARNING BASED SEMANTIC MATCHING FOR TEXTUAL SAMPLES

摘要

#$%^&*AU2018278988A120190117.pdf#####ABSTRACT A device, including one or more processors that receive an input sample of textual content, identify one or more comparison samples that are semantically similar to the input sample, the one or more comparison samples identified based on one or more similarity scores, of the one or more comparison samples and the input sample, satisfying a semantic similarity threshold, identify a plurality of output samples of textual content based on a model, the model receiving, as input, the input sample and acceptance information corresponding to the plurality of output samples and the one or more comparison samples, the model outputting information that identifies the plurality of output samples, the acceptance information determined based on a user input regarding similarity or relevance of the plurality of output samples to the one or more comparison samples, the user input received before the input sample is received, iteratively update the model based on the acceptance information and the identified plurality of output samples, and store or provide the plurality of output samples.1/12 o x CL U) In cY E 0 -0 (U OE cE u E 00 CL a) 4J CL. 4-j 'A - (0 LL (U x 0 0 o - z