Hotspot information marking method and system based on pre-trained intelligent label model

• Main Authors: WANG YING, WEI YONGQIANG, XIA GUANGYU
• Format: Patent
• Language: Chinese; English
• Published: 02.04.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention provides a hotspot information marking method and system based on a pre-trained smart label model, and the method comprises the steps: obtaining to-be-marked text data of a target item, and carrying out the preprocessing of the to-be-marked text data, and obtaining input data; inputting the input data into the pre-trained intelligent label model, and performing fine tuning on the intelligent label model based on the characteristic information of the target item; performing feature extraction on the input data through the fine-tuned intelligent label model to obtain feature representation data; and based on a multi-head attention mechanism, marking and classifying the feature representation data through the finely-adjusted intelligent label model so as to mark hotspot information in the input data. According to the method, the hot spot information can be automatically marked in the related file of the project, and the hot spot marking accuracy and efficiency are improved. 本申请提出了一种基于预训练智能标签模型的热点信息