DATA PROCESSING METHOD AND PRONOUN RESOLUTION NEURAL NETWORK TRAINING METHOD

摘要

The present application relates to a data processing method and a pronoun resolution neural network training method. The method comprises: a pronoun resolution neural network can make good use of features, in text to be detected, corresponding to a context word sequence and a candidate substitution word, performs positive example iteration processing and negative example iteration processing on the features corresponding to the context word sequence and the candidate substitution word so as to obtain a corresponding positive example feature vector modulus length and a corresponding negative example feature vector modulus length, and finally performs calculation according to the positive example feature vector modulus length and the negative example feature vector modulus length so as to obtain the probability of substitution corresponding to each candidate substitution word in a candidate substitution word set.