Petroleum Exploration and Development Text Triplet Extraction Based on Deep Learning

The petroleum exploration and development industry is moving from digital to intelligent. Under the guidance of AI, machine reading and automatic extraction of petroleum exploration and development knowledge are needed for unstructured datas. There are a large number of long entities and complex nested entities in petroleum exploration and development text, which increase the challenge of petroleum exploration and development triplet extraction task. To solve the above problems, (1) The ALBERT-BiLSTM-Attention-CRF method based on large-scale pre-trained Chinese language model is used to extract text triples for Petroleum exploration and development. (2) The ALBERT-BiGRU-Attention method is used to carry out text dichotomies to judge whether triplet extraction is effective. By collecting the datas of Petroleum exploration and development, the corpus of Petroleum exploration and development is established. Secondly, knowledge composition is analyzed and corpus is annotated under the guidance of Petroleum exploration and development theory. Finally, the deep learning training of Petroleum exploration and development descriptive corpus knowledge extraction is carried out. The experimental results show that the accuracy of SPO triplet named entity recognition is 85.49%. The recognition accuracy of extracted triples is 95.90%, which can achieve good recognition effect in small-scale Petroleum exploration and development corpus.