基于深度学习的关系抽取系统实现

The implementation of relationship extraction system based on deep learning

近年来,关系抽取任务已经成为自然语言处理领域的研究热点。本文采用远监督方法,设计一套适用于各领域关系列表的数据集信息抽取方法,通过基于膨胀的卷积神经网络的命名实体识别方法、基于门控循环单元及注意力机制的关系抽取方法来完成基于深度学习的信息抽取方法。通过结果分析,证明本文设计的信息抽取方法有效。

In recent years,the task of relation extraction has become a hot topic in the field of naturallanguageprocessing.In this paper,a set of data set information extraction methods are designed by far supervision method,which is suitable for the list of relationships in various fields.The method of named entity recognition based on convolutional neural network,relation extraction method based on gate control cycle unit and attention mechanism is used to complete the information extraction method based on deep learning.The results show that the information extraction method designed in this paper is effective.