Mining system for Chinese-Uighur-Chinese institution name dictionary based on neural network
The invention discloses a mining system for a Chinese-Uighur-Chinese institution name dictionary based on a neural network. The mining system is composed of an institution name classifier based on a TextCNN model and an LSTM model andan organization name translation model based on a moses model, a T...
Gespeichert in:
Hauptverfasser: | , , , , , , , |
---|---|
Format: | Patent |
Sprache: | chi ; eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | The invention discloses a mining system for a Chinese-Uighur-Chinese institution name dictionary based on a neural network. The mining system is composed of an institution name classifier based on a TextCNN model and an LSTM model andan organization name translation model based on a moses model, a Transformer model and a GNMT model. Fusing the optimal results, firstly identifying organization names of sentences, then selecting whether the sentences are in a dictionary or not, then classifying the sentences by utilizing a classifier, and finally reversely translating the sentences by utilizinga translation model to obtain bilingual organization names and adding the bilingual organization names into the dictionary. The invention provides a translation model with better reverse translation,and a system for continuously mining as many organization name dictionaries as possible and with high quality. The system is a translation model mainly based on a dictionary, classification and a neural network. By comparing d |
---|