Attention-Based Relation Extraction With Bidirectional Gated Recurrent Unit and Highway Network in the Analysis of Geological Data

Abstract

Attention-based deep learning model as a human-centered smart technology has become the state-of-the-art method in addressing relation extraction, while implementing natural language processing. How to effectively improve the computational performance of that model has always been a research focus in both academic and industrial communities. Generally, the structures of model would greatly affect the final results of relation extraction. In this article, a deep learning model with a novel structure is proposed. In our model, after incorporating the highway network into a bidirectional gated recurrent unit, the attention mechanism is additionally utilized in an effort to assign weights of key issues in the network structure. Here, the introduction of highway network could enable the proposed model to capture much more semantic information. Experiments on a popular benchmark data set are conducted, and the results demonstrate that the proposed model outperforms some existing relation extraction methods. Furthermore, the performance of our method is also tested in the analysis of geological data, where the relation extraction in Chinese geological field is addressed and a satisfactory display result is achieved.

Keywords

• Relation extraction
• bidirectional gated recurrent unit (BGRU)
• highway network
• attention
• geological data