Science popularization field named entity identification method, equipment and medium

• Main Authors: SHAO QUANSEN, FENG SHIXIANG, LIU HONGCHEN, LIANG WEICHAO, SU JUNSHUN, GUO YAN, LIN WEIHUA, LIU FUJIANG
• Format: Patent
• Language: Chinese; English
• Published: 01.03.2024
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention provides a science popularization field named entity recognition method, and relates to the field of named entity recognition, and the method comprises the steps: carrying out the preprocessing of a corpus of a to-be-recognized field, generating a field corpus, carrying out the Chinese word segmentation of the field corpus, and determining the word frequency weight sequence representation; sampling the named entity training data set for data enhancement, and determining text input of the model; performing Chinese word segmentation, mapping and splicing on the text input to obtain domain feature input; performing context feature fusion on domain feature input by using a multi-head attention mechanism, and outputting a vocabulary-level feature vector; based on the MRC task, constructing input, and determining character-level vector representation; the character-level vector representation and the vocabulary-level feature vectors are spliced and input into a classifier, and output is prediction out