Fusion Deep Learning and Machine Learning for Heterogeneous Military Entity Recognition

Abstract

With respect to the fuzzy boundaries of military heterogeneous entities, this paper improves the entity annotation mechanism for entity with fuzzy boundaries based on related research works. This paper applies a BERT-BiLSTM-CRF model fusing deep learning and machine learning to recognize military entities, and thus, we can construct a smart military knowledge base with these entities. Furthermore, we can explore many military AI applications with the knowledge base and military Internet of Things (MIoT). To verify the performance of the model, we design multiple types of experiments. Experimental results show that the recognition performance of the model keeps improving with the increasing size of the corpus in the multidata source scenario, with the F -score increasing from 73.56% to 84.53%. Experimental results of cross-corpus cross-validation show that the more types of entities covered in the training corpus and the richer the representation type, the stronger the generalization ability of the trained model, in which the recall rate of the model trained with the novel random type corpus reaches 74.33% and the F -score reaches 76.98%. The results of the multimodel comparison experiments show that the BERT-BiLSTM-CRF model applied in this paper performs well for the recognition of military entities. The longitudinal comparison experimental results show that the F -score of the BERT-BiLSTM-CRF model is 18.72%, 11.24%, 9.24%, and 5.07% higher than the four models CRF, LSTM-CRF, BiLSTM-CR, and BERT-CRF, respectively. The cross-sectional comparison experimental results show that the F -score of the BERT-BiLSTM-CRF model improved by 6.63%, 7.95%, 3.72%, and 1.81% compared to the Lattice-LSTM-CRF, CNN-BiLSTM-CRF, BERT-BiGRU-CRF, and BERT-IDCNN-CRF models, respectively.

Surgical procedure long terms recognition from Chinese literature incorporating structural feature

“…In this work, we conduct several comprehensive experiments to verify the proposed module. The comparison models include (1) BERT-Bi-LSTM-CRF [ 61 ], a BERT model with a bidirectional long-short term memory network (LSTM) followed by a CRF layer that is the most popular architecture for NER task, but it still lacks the ability to leverage lexicon information; (2) BERT-FLAT-Lattice-Transformer [ 50 ], a BERT model with a flat-lattice Transformer incorporating lexicon information, but it is too complicated in structure and is hard to deploy just in embedding layer; (4) SSBC, our proposed model simply fuse lexicon-character level feature without pre-trained language model; and (4) Bichar-SSBC, our proposed model with bichar feature embedding. The model comparison, in theory, can prove the importance of lexicon information and a pre-trained language model.…”

MLRIP: Pre-training a military language representation model with informative factual knowledge and professional knowledge base

“…In this proposed work, we choose five categories of professional data, namely operational documents, intelligence documents, military scenarios, military books and military simulation system logs, and seven categories of network data, namely military website documents, military game strategies, military news, military comments, military blogs, military forum data, and Chinese Wikipedia as our training corpora. We have analyzed and categorized this data in detail in our previous work (Li et al, 2022), and refer readers to that work for the data information. Our previous work shows that the word distributions are quite different in different data source texts, for example, military terms are mostly expressed in formal or standard expressions in military books, while military terms are dominated by abbreviations and code names in military intelligence texts.…”

Intelligent generation algorithm for fusion model based on massive multi-source heterogeneous data

“…Data fusion is usually divided into three levels: data level, feature level, and decision level, with the fusion level ranking from low to high. Usually, the data level is used to fuse sensor data of the same type, the feature level is used to fuse heterogeneous sensor data, and the decision level obtains the final evaluation result by fusing multi-source data [9,10]. Based on this, this study proposes a hybrid data fusion algorithm based on multi-source heterogeneous sensor data for building structural characteristics, which is data-driven to achieve engineering operation safety status evaluation.…”