Chinese named entity recognition(CNER)has received widespread attention as an important task of Chinese information extraction.Most previous research has focused on individually studying flat CNER,overlapped CNER,or d...Chinese named entity recognition(CNER)has received widespread attention as an important task of Chinese information extraction.Most previous research has focused on individually studying flat CNER,overlapped CNER,or discontinuous CNER.However,a unified CNER is often needed in real-world scenarios.Recent studies have shown that grid tagging-based methods based on character-pair relationship classification hold great potential for achieving unified NER.Nevertheless,how to enrich Chinese character-pair grid representations and capture deeper dependencies between character pairs to improve entity recognition performance remains an unresolved challenge.In this study,we enhance the character-pair grid representation by incorporating both local and global information.Significantly,we introduce a new approach by considering the character-pair grid representation matrix as a specialized image,converting the classification of character-pair relationships into a pixel-level semantic segmentation task.We devise a U-shaped network to extract multi-scale and deeper semantic information from the grid image,allowing for a more comprehensive understanding of associative features between character pairs.This approach leads to improved accuracy in predicting their relationships,ultimately enhancing entity recognition performance.We conducted experiments on two public CNER datasets in the biomedical domain,namely CMeEE-V2 and Diakg.The results demonstrate the effectiveness of our approach,which achieves F1-score improvements of 7.29 percentage points and 1.64 percentage points compared to the current state-of-the-art(SOTA)models,respectively.展开更多
6 Atomic fragment types of organic compound have been defined, and the multilevel atom-pair frequency matrix has been constructed according to the occurrence number in pairs of atomic fragments with different bond len...6 Atomic fragment types of organic compound have been defined, and the multilevel atom-pair frequency matrix has been constructed according to the occurrence number in pairs of atomic fragments with different bond lengths in the molecule. On the basis of them, a novel molecular coding technique： characteristic atom-pair holographic code （CAHC）, is obtained. To some extent, this method exhibits a large number of benefits at the same time. For example, it can calculate 2D molecular topological descriptor easily, operate without difficulty and possess definite physicochemical meaning of 3D molecular structural characterization methods, and may fetch the complicated information of molecule, etc. Therefore, it is appropriate for the study on quantitative structure-property/activity relationship （QSPR/QSAR） of medicines and biological molecules. We attempt in this paper to utilize the method of CAHC to the quantitative prediction of reversed-phase liquid chromatogram （RPLC） retention data of 33 purine derivatives and 24 steroids. The fitting multiple correlation coefficient R2, cross-validated multiple correlation coefficient Q2 and predicted ability Q^2 pred over test set＇s samples of obtained partial least-square （PLS） regression model are respectively 0.990, 0.893 and 0.977, 0.897, 0.941.展开更多
基金supported by Yunnan Provincial Major Science and Technology Special Plan Projects(Grant Nos.202202AD080003,202202AE090008,202202AD080004,202302AD080003)National Natural Science Foundation of China(Grant Nos.U21B2027,62266027,62266028,62266025)Yunnan Province Young and Middle-Aged Academic and Technical Leaders Reserve Talent Program(Grant No.202305AC160063).
文摘Chinese named entity recognition(CNER)has received widespread attention as an important task of Chinese information extraction.Most previous research has focused on individually studying flat CNER,overlapped CNER,or discontinuous CNER.However,a unified CNER is often needed in real-world scenarios.Recent studies have shown that grid tagging-based methods based on character-pair relationship classification hold great potential for achieving unified NER.Nevertheless,how to enrich Chinese character-pair grid representations and capture deeper dependencies between character pairs to improve entity recognition performance remains an unresolved challenge.In this study,we enhance the character-pair grid representation by incorporating both local and global information.Significantly,we introduce a new approach by considering the character-pair grid representation matrix as a specialized image,converting the classification of character-pair relationships into a pixel-level semantic segmentation task.We devise a U-shaped network to extract multi-scale and deeper semantic information from the grid image,allowing for a more comprehensive understanding of associative features between character pairs.This approach leads to improved accuracy in predicting their relationships,ultimately enhancing entity recognition performance.We conducted experiments on two public CNER datasets in the biomedical domain,namely CMeEE-V2 and Diakg.The results demonstrate the effectiveness of our approach,which achieves F1-score improvements of 7.29 percentage points and 1.64 percentage points compared to the current state-of-the-art(SOTA)models,respectively.
基金This work was supported by the State Key Laboratory of Chemo/Biosensing and Chemometrics Foundation (No. 05-12-1), Fok-Yingtung Educational Foundation (No. 98-7-6) and Chongqing University Innovation Foundation of Science and Technology ( No. 06-1-1)
文摘6 Atomic fragment types of organic compound have been defined, and the multilevel atom-pair frequency matrix has been constructed according to the occurrence number in pairs of atomic fragments with different bond lengths in the molecule. On the basis of them, a novel molecular coding technique： characteristic atom-pair holographic code （CAHC）, is obtained. To some extent, this method exhibits a large number of benefits at the same time. For example, it can calculate 2D molecular topological descriptor easily, operate without difficulty and possess definite physicochemical meaning of 3D molecular structural characterization methods, and may fetch the complicated information of molecule, etc. Therefore, it is appropriate for the study on quantitative structure-property/activity relationship （QSPR/QSAR） of medicines and biological molecules. We attempt in this paper to utilize the method of CAHC to the quantitative prediction of reversed-phase liquid chromatogram （RPLC） retention data of 33 purine derivatives and 24 steroids. The fitting multiple correlation coefficient R2, cross-validated multiple correlation coefficient Q2 and predicted ability Q^2 pred over test set＇s samples of obtained partial least-square （PLS） regression model are respectively 0.990, 0.893 and 0.977, 0.897, 0.941.
