A sentence backbone analysis method and system based on multi-layer error feedback neural network of word segmentation and named entity recognition

• Main Authors: CHEN TAO, WU MINGFEN
• Format: Patent
• Language: Chinese; English
• Published: 18.01.2019
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention provides a sentence backbone analysis method and system based on multi-layer error feedback neural network of word segmentation and named entity recognition.The invention firstly segments Chinese sentences into word sequences, then identifies the named entity of the word sequences, merges the words belonging to the same named entity, finally analyzes the backbone of the recognized sentences of the named entity, and visually outputs the main components of the recognized sentences. The invention adopts an artificial neural network based on depth learning and a method of combining multi-layer semantic element structure information from a word to a named entity and then to a sentence, respectively trains and optimizes the depth neural network according to the structure information of different layers, and improves the effect of sentence backbone analysis through multi-layer error feedback. The method can improve the system accuracy, response speed and fault tolerance. 本发明提供种基于分词和命名实体识别的多层误差反馈神经网络的句子