Filters








21 Hits in 0.72 sec

Element Intervention for Open Relation Extraction [article]

Fangchao Liu, Lingyong Yan, Hongyu Lin, Xianpei Han, Le Sun
2021 arXiv   pre-print
Open relation extraction aims to cluster relation instances referring to the same underlying relation, which is a critical step for general relation extraction. Current OpenRE models are commonly trained on the datasets generated from distant supervision, which often results in instability and makes the model easily collapsed. In this paper, we revisit the procedure of OpenRE from a causal view. By formulating OpenRE using a structural causal model, we identify that the above-mentioned problems
more » ... stem from the spurious correlations from entities and context to the relation type. To address this issue, we conduct Element Intervention, which intervenes on the context and entities respectively to obtain the underlying causal effects of them. We also provide two specific implementations of the interventions based on entity ranking and context contrasting. Experimental results on unsupervised relation extraction datasets show that our methods outperform previous state-of-the-art methods and are robust across different datasets.
arXiv:2106.09558v1 fatcat:pairkdghknc53ajnsbryorqdb4

Superconductivity in CuIr2-xAlxTe4 telluride chalcogenides [article]

Dong Yan, Lingyong Zeng, Yijie Zeng, Yishi Lin, Junjie Yin, Meng Wang, Yihua Wang, Dao-Xin Yao, Huixia Luo
2021 arXiv   pre-print
Chinese Physics B Superconductivity in CuIr2-xAlxTe4 telluride chalcogenides Dong Yan (严冬)1,4 #, Lingyong Zeng (曾令勇)1 #, Yijie Zeng (曾宜杰)2,5, Yishi Lin (林一石)3, Junjie Yin (殷俊杰)2, Meng Wang  ...  Hu W Z, Li G, Yan J, Wen H H, Wu G, Chen X H and Wang N L 2007 Phys. Rev. B 76 045103 31. Yan D, Wang S, Lin Y S, Wang G H, Zeng Y J, Boubeche M, He Y, Ma J, Wang Y H, Yao D X Luo H X 2020 J.  ... 
arXiv:2112.11615v1 fatcat:jluurehdmjdqfpsorqbn73ek2a

Progressive Adversarial Learning for Bootstrapping: A Case Study on Entity Set Expansion [article]

Lingyong Yan, Xianpei Han, Le Sun
2021 arXiv   pre-print
For example, Yan et al. (2020a) propose an end-toend bootstrapping network learned by multi-view learning, and extend it by self-supervised and supervised pre-training (Yan et al., 2020b) .  ...  et al., 2020a) , end-to-end bootstrapping model pre-trained using external datasets-GBN (Yan et al., 2020b) .  ... 
arXiv:2109.12082v1 fatcat:inuocx3sqrfkddv3lmtwv4jmuq

Knowledgeable or Educated Guess? Revisiting Language Models as Knowledge Bases [article]

Boxi Cao, Hongyu Lin, Xianpei Han, Le Sun, Lingyong Yan, Meng Liao, Tong Xue, Jin Xu
2021 arXiv   pre-print
Previous literatures show that pre-trained masked language models (MLMs) such as BERT can achieve competitive factual knowledge extraction performance on some datasets, indicating that MLMs can potentially be a reliable knowledge source. In this paper, we conduct a rigorous study to explore the underlying predicting mechanisms of MLMs over different extraction paradigms. By investigating the behaviors of MLMs, we find that previous decent performance mainly owes to the biased prompts which
more » ... it dataset artifacts. Furthermore, incorporating illustrative cases and external contexts improve knowledge prediction mainly due to entity type guidance and golden answer leakage. Our findings shed light on the underlying predicting mechanisms of MLMs, and strongly question the previous conclusion that current MLMs can potentially serve as reliable factual knowledge bases.
arXiv:2106.09231v1 fatcat:kmfjzrpq65hf3lc5h23x2rwsge

Investigation of heavy oil refinery wastewater treatment by integrated ozone and activated carbon -supported manganese oxides

Chunmao Chen, Lingyong Wei, Xuan Guo, Shaohui Guo, Guangxu Yan
2014 Fuel processing technology  
doi:10.1016/j.fuproc.2014.02.024 fatcat:ev3eun5x5vd7hb6prq4a7fmgja

From Bag of Sentences to Document: Distantly Supervised Relation Extraction via Machine Reading Comprehension [article]

Lingyong Yan, Xianpei Han, Le Sun, Fangchao Liu, Ning Bian
2020 arXiv   pre-print
Distant supervision (DS) is a promising approach for relation extraction but often suffers from the noisy label problem. Traditional DS methods usually represent an entity pair as a bag of sentences and denoise labels using multi-instance learning techniques. The bag-based paradigm, however, fails to leverage the inter-sentence-level and the entity-level evidence for relation extraction, and their denoising algorithms are often specialized and complicated. In this paper, we propose a new DS
more » ... digm–document-based distant supervision, which models relation extraction as a document-based machine reading comprehension (MRC) task. By re-organizing all sentences about an entity as a document and extracting relations via querying the document with relation-specific questions, the document-based DS paradigm can simultaneously encode and exploit all sentence-level, inter-sentence-level, and entity-level evidence. Furthermore, we design a new loss function–DSLoss (distant supervision loss), which can effectively train MRC models using only ⟨document, question, answer⟩ tuples, therefore noisy label problem can be inherently resolved. Experiments show that our method achieves new state-of-the-art DS performance.
arXiv:2012.04334v2 fatcat:4ab72nh2vfdrtddlrg5qienc2u

Superconductivity in Ru-doped CuIr2Te4 telluride chalcogenide

Dong Yan, Lingyong Zeng, Yishi Lin, Junjie Yin, Yuan He, Xing Zhang, Meiling Huang, Bing Shen, Meng Wang, Yihua Wang, Daoxin Yao, Huixia Luo
2019 Physical review B  
Here we report the effect of structural and superconductivity properties on Ru doped CuIr2Te4 telluride chalcogenide. XRD results suggest that the CuIr2-xRuxTe4 maintain the disordered trigonal structure with space group P3m1 (No. 164) for x less than 0.3. The lattice constants, a and c, both decrease with increasing Ru content. Temperature-dependent resistivity, magnetic susceptibility and specific-heat measurements are performed to characterize the superconducting properties systematically.
more » ... r results suggest that the optimal doping level for superconductivity in CuIr2-xRuxTe4 is x = 0.05, where Tc is 2.79 K with the Sommerfeld constant gamma of 11.52 mJ mol-1 K-2 and the specific-heat anomaly at the superconducting transition, is approximately 1.51, which is higher than the BCS value of 1.43, indicating CuIr1.95Ru0.05Te4 is a strongly electron-phonon coupled superconductor. The values of lower critical filed and upper critical field calculated from isothermal magnetization and magneto-transport measurements are 0.98 KOe and 2.47 KOe respectively, signifying that the compound is clearly a type-II superconductor. Finally, a dome-like shape superconducting Tcs vs. x content phase diagram is established, where the charge density wave disappears at x = 0.03 while superconducting transition temperature (Tc) rises until it reaches its peak at x = 0.05, then, with decreasing when x reaches 0.3. This feature of the competition between CDW and the superconductivity could be caused by tuning the Fermi surface and density of states with Ru chemical doping.
doi:10.1103/physrevb.100.174504 fatcat:ot7cedkqbzfifepmh7xgn6ni3q

Learning to Bootstrap for Entity Set Expansion

Lingyong Yan, Xianpei Han, Le Sun, Ben He
2019 Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)  
Bootstrapping for Entity Set Expansion (ESE) aims at iteratively acquiring new instances of a specific target category. Traditional bootstrapping methods often suffer from two problems: 1) delayed feedback, i.e., the pattern evaluation relies on both its direct extraction quality and the extraction quality in later iterations. 2) sparse supervision, i.e., only few seed entities are used as the supervision. To address the above two problems, we propose a novel bootstrapping method combining the
more » ... onte Carlo Tree Search (MCTS) algorithm with a deep similarity network, which can efficiently estimate delayed feedback for pattern evaluation and adaptively score entities given sparse supervision signals. Experimental results confirm the effectiveness of the proposed method.
doi:10.18653/v1/d19-1028 dblp:conf/emnlp/YanHSH19 fatcat:yhb3i7kylretfaz65bcvxukhia

Charge density wave and superconductivity in the family of telluride chalcogenides Zn1-xCuxIr2-yN(N = Al, Ti, Rh)yTe4 [article]

Dong Yan, Yijie Zeng, Yishi Lin, Lingyong Zeng, Junjie Yin, Yuan He, Mebrouka Boubeche, Meng Wang, Yihua Wang, Dao-Xin Yao, Huixia Luo
2020 arXiv   pre-print
Supporting Information Charge density wave and superconductivity in the family of telluride chalcogenides Zn1-xCuxIr2-yN(N = Al, Ti, Rh)yTe4 Dong Yan a , Yijie Zeng b , Yishi Lin c , Lingyong Zeng a ,  ... 
arXiv:2003.11463v1 fatcat:knphfgn7cbaknfjmlcgdjjfjwu

End-to-End Bootstrapping Neural Network for Entity Set Expansion

Lingyong Yan, Xianpei Han, Ben He, Le Sun
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
paradigm, the expanded entities from previous steps are directly used as the golden entities to guide the next expansion (Riloff and Jones 1999; Curran, Murphy, and Scholz 2007; Gupta and Manning 2014; Yan  ...  (Zupon et al. 2019) : this method follows Gupta and Manning (2014) , but learns custom embeddings for entities and patterns at each iteration, which are used to guide the entity classifier. • LTB (Yan  ... 
doi:10.1609/aaai.v34i05.6482 fatcat:5traskqpq5ev5m4ndrjngivhwq

Exploration of the Peptide Recognition of an Amiloride-sensitive FMRFamide Peptide-gated Sodium Channel

You-Ya Niu, Yang Yang, Yan Liu, Li-Dong Huang, Xiao-Na Yang, Ying-Zhe Fan, Xiao-Yang Cheng, Peng Cao, You-Min Hu, Lingyong Li, Xiang-Yang Lu, Yun Tian (+1 others)
2016 Journal of Biological Chemistry  
FMRFamide (Phe-Met-Arg-Phe-NH 2 )-activated sodium channel (FaNaC) is an amiloride-sensitive sodium channel activated by endogenous tetrapeptide in invertebrates, and belongs to the epithelial sodium channel/degenerin (ENaC/DEG) superfamily. The ENaC/DEG superfamily differs markedly in its means of activation, such as spontaneously opening or gating by mechanical stimuli or tissue acidosis. Recently, it has been observed that a number of ENaC/DEG channels can be activated by small molecules or
more » ... eptides, indicating that the ligand-gating may be an important feature of this superfamily. The peptide ligand control of the channel gating might be an ancient ligand-gating feature in this superfamily. Therefore, studying the peptide recognition of FaNaC channels would advance our understanding of the ligand-gating properties of this superfamily of ion channels. Here we demonstrate that Tyr-131, Asn-134, Asp-154, and Ile-160, located in the putative upper finger domain of Helix aspersa FaNaC (HaFaNaC) channels, are key residues for peptide recognition of this ion channel. Two HaFaNaC specificinsertion motifs among the ENaC/DEG superfamily, residing at the putative ␣4-␣5 linker of the upper thumb domain and the ␣6-␣7 linker of the upper knuckle domain, are also essential for the peptide recognition of HaFaNaC channels. Chemical modifications and double mutant cycle analysis further indicated that those two specific inserts and key residues in the upper finger domain together participate in peptide recognition of HaFaNaC channels. This ligand recognition site is distinct from that of acid-sensing ion channels (ASICs) by a longer distance between the recognition site and the channel gate, carrying useful information about the ligand gating and the evolution of the trimeric ENaC/DEG superfamily of ion channels. The epithelial sodium channel/Degenerin (ENaC/DEG) 4 superfamily, cloned in the early 1990s, is nonvoltage-gated, amiloride-sensitive, and sodium-selective ion channels, including ASIC, ENaC, DEG, FaNaC, bile acid-sensitive ion channel (BASIC), and pickpocket (PPK) channels, etc. (1, 2) . ENaC/DEG ion channels are implicated in many physiological and pathological functions such as synaptic plasticity, learning and memory, emotion regulation, neurodegenerative diseases, epileptic seizures, pain sensation, mechano-sensation, blood pressure regulation, and cystic fibrosis (1-3), which makes them potential drug targets for those disorders. The members of the ENaC/ DEG superfamily differ markedly by means of activation. For example, DEG channels are mechano-sensitive; ENaC channels open spontaneously; ASIC channels are capable of sensing tissue acidosis, while FaNaC is activated by RFamide peptides, etc. (4). Recently, great advances have been made in the exploration of the activation mechanism of this superfamily. For example, it is now known that ENaC, ASIC3, BASIC, and ASIC1a channels can be activated by the small molecules S3969 (5), GMQ (6), bile acid (7) , and peptide toxin (8, 9) respectively. These new findings suggest an important small molecule or peptide ligand-gating property of this superfamily of ion channels, in addition to the subtype-specific gating mechanisms such as spontaneously opening or gating by mechanical stimuli and tissue acidosis, can play an important role in ENaC/DEG channel activation. However, the structural components underlying the ligand gating of ENaC/DEG remain largely unclear except for ASIC channels, mainly due to the high resolution structure of chicken ASIC1 (cASIC1) that was determined recently. However, extensively studying gating properties of ASIC channels remains difficult because of multiple proton-binding sites located in different domains of ASIC channels (10 -13). . B, representative current traces of ⌬445-493 induced by different concentrations of FMRFamide. C, pooled data showing ratio of 7 M and 100 M FMRFamide triggered currents of the mutants in HaFaNaC specific-insertion II. Data are presented as means Ϯ S.E. (n ϭ 4). ## , p Ͻ 0.01; ** , p Ͻ 0.01 versus WT, Student's t test. D, representative images showing Western blots of total and cell surface expressions of HaFaNaC WT channels and ⌬445-493 mutants. E, dose-response relationship of HaFaNaC WT and mutants in HaFaNaC specific-insertion II. Data points are means Ϯ S.E. (n ϭ 4 -8) and fit with Hill equation. Downloaded from FIGURE 7. Hypothetical activation models for DEG, ASIC and FaNaC. A, DEG model showing ectodomain sensing the mechanical stimulus. B, ASIC model showing proton and GMQ binding sites, respectively. C, putative FMRFamide binding sites in HaFaNaC, * indicates residues from a neighboring subunit.
doi:10.1074/jbc.m115.710251 pmid:26867576 pmcid:PMC4817185 fatcat:ryfxe7xcsjca7dgvxf4zh74yvq

Intersubunit physical couplings fostered by the left flipper domain facilitate channel opening of P2X4 receptors

Jin Wang, Liang-Fei Sun, Wen-Wen Cui, Wen-Shan Zhao, Xue-Fei Ma, Bin Li, Yan Liu, Yang Yang, You-Min Hu, Li-Dong Huang, Xiao-Yang Cheng, Lingyong Li (+3 others)
2017 Journal of Biological Chemistry  
Edited by F. Anne Stephenson P2X receptors are ATP-gated trimeric channels with important roles in diverse pathophysiological functions. A detailed understanding of the mechanism underlying the gating process of these receptors is thus fundamentally important and may open new therapeutic avenues. The left flipper (LF) domain of the P2X receptors is a flexible loop structure, and its coordinated motions together with the dorsal fin (DF) domain are crucial for the channel gating of the P2X
more » ... rs. However, the mechanism underlying the crucial role of the LF domain in the channel gating remains obscure. Here, we propose that the ATP-induced allosteric changes of the LF domain enable it to foster intersubunit physical couplings among the DF and two lower body domains, which are pivotal for the channel gating of P2X4 receptors. Metadynamics analysis indicated that these newly established intersubunit couplings correlate well with the ATP-bound open state of the receptors. Moreover, weakening or strengthening these physical interactions with engineered intersubunit metal bridges remarkably decreased or increased the open probability of the receptors, respectively. Further disulfide cross-linking and covalent modification confirmed that the intersubunit physical couplings among the DF and two lower body domains fostered by the LF domain at the open state act as an integrated structural element that is stringently required for the channel gating of P2X4 receptors. Our observations provide new mechanistic insights into P2X receptor activation and will stimulate development of new allosteric modulators of P2X receptors. P2X receptors are trimeric membrane ion channels (1, 2) activated by extracellular ATP (3). So far, seven P2X subtypes (P2X1-P2X7) have been identified, which are expressed virtu-ally in almost all mammalian tissues, including nervous, immune, and cardiovascular systems (3-5). P2X receptors possess a molecular architecture distinct from other ion channel protein families (4, 6, 7) and are implicated in a wide range of physiological and pathological processes (2, 4, 5, 8), such as neuroinflammation, synaptic transmission, primary afferent signaling, chronic pain, central control of respiration, vascular remodeling, and the regulation of blood pressure. Accordingly, P2X receptors hold great interest as new therapeutic targets for inflammation and v cardiovascular and neurological diseases (8 -14). For this purpose, it is essential to fully understand the detailed gating mechanism of P2X receptors at the atomic level (9, 15, 16) . High resolution X-ray structures at apo/closed and ATPbound open states are available for zebrafish P2X4 (zfP2X4) (7, 9), human P2X3 (hP2X3) (17), Amblyomma maculatum P2X (AmP2X) (18), and panda P2X7 (19) receptors, which greatly aid our understanding of the working principles of those unique receptors, such as the ligand recognition, pore architecture, and conformational changes associated with the channel activation. Based on these structures, optical control of P2X receptors independent of natural stimulus has been achieved via powerful optogating approaches (20 -22). The structural comparison of the closed and open structures suggests a possible gating mechanism of P2X receptors (Fig. 1, A and B) (9, 16, 23, 24) . First, at the ATP-binding site, ATP promotes the jaw closure between the head and dorsal fin (DF) 3 domains, making the DF domain move upward to the head domain to accommodate ATP. Meanwhile, the bound ATP pushes the left flipper (LF) domain out of the ATP-binding site. Second, because both the LF and DF domains are structurally coupled with the lower body domain, the movements of those two domains lead to a concomitant outward flexing of lower body domains in the open state, which markedly expands the central vestibule (Fig. 1B) . Finally, lower body domains are directly coupled with two transmembrane (TM) domains TM1 and TM2, and therefore their outward flexing can directly promote the opening of ion
doi:10.1074/jbc.m116.771121 pmid:28302727 pmcid:PMC5418059 fatcat:6qzqa5zjhng4jlnw3rvpiygsje

A Highly Conserved Salt Bridge Stabilizes the Kinked Conformation of β2,3-Sheet Essential for Channel Function of P2X4 Receptors

Wen-Shan Zhao, Meng-Yang Sun, Liang-Fei Sun, Yan Liu, Yang Yang, Li-Dong Huang, Ying-Zhe Fan, Xiao-Yang Cheng, Peng Cao, You-Min Hu, Lingyong Li, Yun Tian (+2 others)
2016 Journal of Biological Chemistry  
Significant progress has been made in understanding the roles of crucial residues/motifs in the channel function of P2X receptors during the pre-structure era. The recent structural determination of P2X receptors allows us to reevaluate the role of those residues/motifs. Residues Arg-309 and Asp-85 (rat P2X4 numbering) are highly conserved throughout the P2X family and were involved in loss-of-function polymorphism in human P2X receptors. Previous studies proposed that they participated in
more » ... t ATP binding. However, the crystal structure of P2X demonstrated that those two residues form an intersubunit salt bridge located far away from the ATP-binding site. Therefore, it is necessary to reevaluate the role of this salt bridge in P2X receptors. Here, we suggest the crucial role of this structural element both in protein stability and in channel gating rather than direct ATP interaction and channel assembly. Combining mutagenesis, charge swap, and disulfide cross-linking, we revealed the stringent requirement of this salt bridge in normal P2X4 channel function. This salt bridge may contribute to stabilizing the bending conformation of the ␤2,3-sheet that is structurally coupled with this salt bridge and the ␣2-helix. Strongly kinked ␤2,3 is essential for domain-domain interactions between head domain, dorsal fin domain, right flipper domain, and loop ␤7,8 in P2X4 receptors. Disulfide cross-linking with directions opposing or along the bending angle of the ␤2,3-sheet toward the ␣2-helix led to loss-of-function and gainof-function of P2X4 receptors, respectively. Further insertion of amino acids with bulky side chains into the linker between the ␤2,3-sheet or the conformational change of the ␣2-helix, interfering with the kinked conformation of ␤2,3, led to loss-of-function of P2X4 receptors. All these findings provided new insights in understanding the contribution of the salt bridge between Asp-85 and Arg-309 and its structurally coupled ␤2,3-sheet to the function of P2X receptors.
doi:10.1074/jbc.m115.711127 pmid:26865631 pmcid:PMC4825005 fatcat:dn3o5icxrbb5te2p6v62r5lzke

Global Bootstrapping Neural Network for Entity Set Expansion

Lingyong Yan, Xianpei Han, Ben He, Le Sun
2020 Findings of the Association for Computational Linguistics: EMNLP 2020   unpublished
Compared to previous end-to-end model (Yan et al., 2020) , the GBEncoder is different in two aspects: 1).  ...  To capture both the local and global semantics, we follow Yan et al. (2020) and use the entitypattern bipartite graph.  ... 
doi:10.18653/v1/2020.findings-emnlp.331 fatcat:l4yhk3whqbgebmxq7ea3k6wpyq

Progressive Adversarial Learning for Bootstrapping: A Case Study on Entity Set Expansion

Lingyong Yan, Xianpei Han, Le Sun
2021 Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing   unpublished
For example, Yan et al. (2020a) propose an end-toend bootstrapping network learned by multi-view learning, and extend it by self-supervised and supervised pre-training (Yan et al., 2020b) .  ...  et al., 2020a) , end-to-end bootstrapping model pre-trained using external datasets-GBN (Yan et al., 2020b) .  ... 
doi:10.18653/v1/2021.emnlp-main.762 fatcat:o65k6oryuvc4pmf6muihzwr5ve
« Previous Showing results 1 — 15 out of 21 results