Deep learning-based text knowledge classification for whole-process engineering consulting standards
Gu Jianan,Ren Kehao,Gao Binwei
DOI: https://doi.org/10.1016/j.jer.2023.07.011
IF: 1.325
2023-07-01
Journal of Engineering Research
Abstract:The knowledge classification technology has significant implications for the intelligent research of industries. In the field of whole-process engineering consulting, manually reading and processing large amounts of text data is both time-consuming and laborious. Knowledge classification technology can automatically classify these text data and extract key information, which can improve industry work efficiency. In this study, a deep learning-based text knowledge classification method is proposed to address the large-scale text classification problem in the whole-process engineering consulting field. Firstly, pre-trained language models such as RoBERTa, BERT, and Longformer-RoBERTa are used to extract features from text. Secondly, a multi-label classification model is used to classify the text. Experimental results show that the proposed method performs better than other commonly used models in both overall classification performance and individual category classification performance. Moreover, when the text knowledge classification model is integrated as a text representation module with common classification models such as CNN and LSTM, its performance is inferior to that of a pure classification model. The proposed text knowledge classification method is of great significance for the application in the field of whole-process engineering consulting and provides an effective solution for intelligent research in engineering consulting.
engineering, multidisciplinary