Refine
Year of publication
Document Type
- Preprint (759)
- Article (402)
- Working Paper (119)
- Doctoral Thesis (93)
- Diploma Thesis (47)
- Conference Proceeding (41)
- Book (37)
- Bachelor Thesis (36)
- diplomthesis (28)
- Report (25)
Has Fulltext
- yes (1619)
Is part of the Bibliography
- no (1619)
Keywords
Institute
- Informatik (1619) (remove)
The azimuthal correlations of D mesons and charged particles were measured with the ALICE detector in pp collisions at s√=7 TeV and p-Pb collisions at sNN−−−√=5.02 TeV at the Large Hadron Collider. D0, D+, and D∗+ mesons and their charge conjugates with transverse momentum 3<pT<16 GeV/c and rapidity in the nucleon-nucleon centre-of-mass system |ycms|<0.5 (pp collisions) and −0.96<ycms<0.04 (p-Pb collisions) were correlated to charged particles with pT>0.3 Gev/c. The properties of the correlation peak induced by the jet containing the D meson, described in terms of the yield of charged particles in the peak and peak width, are compatible within uncertainties between the two collision systems, and described by Monte-Carlo simulations based on the PYTHIA, POWHEG and EPOS 3 event generators.
The azimuthal correlations of D mesons and charged particles were measured with the ALICE detector in pp collisions at s√=7 TeV and p-Pb collisions at sNN−−−√=5.02 TeV at the Large Hadron Collider. D0, D+, and D∗+ mesons with transverse momentum 3<pT<16 GeV/c and rapidity in the nucleon-nucleon centre-of-mass system |ycms|<0.5 (pp collisions) and −0.96<ycms<0.04 (p-Pb collisions) were correlated to charged particles with pT>0.3 Gev/c. The properties of the correlation peak induced by the jet containing the D meson, described in terms of the yield of charged particles in the peak and peak width, are compatible within uncertainties between the two collision systems, and described by Monte-Carlo simulations based on the PYTHIA and POWHEG event generators.
Measurement of anti-3He nuclei absorption in matter and impact on their propagation in the Galaxy
(2022)
In our Galaxy, light antinuclei composed of antiprotons and antineutrons can be produced through high-energy cosmic-ray collisions with the interstellar medium or could also originate from the annihilation of dark-matter particles that have not yet been discovered. On Earth, the only way to produce and study antinuclei with high precision is to create them at high-energy particle accelerators. Although the properties of elementary antiparticles have been studied in detail, the knowledge of the interaction of light antinuclei with matter is limited. We determine the disappearance probability of 3He when it encounters matter particles and annihilates or disintegrates within the ALICE detector at the Large Hadron Collider. We extract the inelastic interaction cross section, which is then used as an input to the calculations of the transparency of our Galaxy to the propagation of 3He stemming from dark-matter annihilation and cosmic-ray interactions within the interstellar medium. For a specifc dark-matter profle, we estimate a transparency of about 50%, whereas it varies with increasing 3He momentum from 25% to 90% for cosmic-ray sources. The results indicate that 3He nuclei can travel long distances in the Galaxy, and can be used to study cosmic-ray interactions and dark-matter annihilation.
Measurement of an excess in the yield of J/ψ at very low pT in Pb–Pb collisions at √sNN = 2.76 TeV
(2016)
We report on the first measurement of an excess in the yield of J/ψ at very low transverse momentum (pT<0.3 GeV/c) in peripheral hadronic Pb-Pb collisions at sNN−−−√ = 2.76 TeV, performed by ALICE at the CERN LHC. Remarkably, the measured nuclear modification factor of J/ψ in the rapidity range 2.5<y<4 reaches about 7 (2) in the pT range 0-0.3 GeV/c in the 70-90% (50-70%) centrality class. The J/ψ production cross section associated with the observed excess is obtained under the hypothesis that coherent photoproduction of J/ψ is the underlying physics mechanism. If confirmed, the observation of J/ψ coherent photoproduction in Pb-Pb collisions at impact parameters smaller than twice the nuclear radius opens new theoretical and experimental challenges and opportunities. In particular, coherent photoproduction accompanying hadronic collisions may provide insight into the dynamics of photoproduction and nuclear reactions, as well as become a novel probe of the Quark-Gluon Plasma.
Measurement of an excess in the yield of J/ψ at very low pT in Pb–Pb collisions at √sNN = 2.76 TeV
(2016)
We report on the first measurement of an excess in the yield of J/ψ at very low transverse momentum (pT<0.3 GeV/c) in peripheral hadronic Pb-Pb collisions at sNN−−−√ = 2.76 TeV, performed by ALICE at the CERN LHC. Remarkably, the measured nuclear modification factor of J/ψ in the rapidity range 2.5<y<4 reaches about 7 (2) in the pT range 0-0.3 GeV/c in the 70-90% (50-70%) centrality class. The J/ψ production cross section associated with the observed excess is obtained under the hypothesis that coherent photoproduction of J/ψ is the underlying physics mechanism. If confirmed, the observation of J/ψ coherent photoproduction in Pb-Pb collisions at impact parameters smaller than twice the nuclear radius opens new theoretical and experimental challenges and opportunities. In particular, coherent photoproduction accompanying hadronic collisions may provide insight into the dynamics of photoproduction and nuclear reactions, as well as become a novel probe of the Quark-Gluon Plasma.
Measurement of an excess in the yield of J/ψ at very low pT in Pb–Pb collisions at √sNN = 2.76 TeV
(2015)
We report on the first measurement of an excess in the yield of J/ψ at very low transverse momentum (pT<0.3 GeV/c) in peripheral hadronic Pb-Pb collisions at sNN−−−√ = 2.76 TeV, performed by ALICE at the CERN LHC. Remarkably, the measured nuclear modification factor (RAA) of J/ψ in the rapidity range 2.5<y<4 reaches about 7 (2) in the pT range 0-0.3 GeV/c in the 70-90% (50-70%) centrality class. The J/ψ production cross section associated with the observed excess is obtained under the hypothesis that coherent photoproduction of J/ψ is the underlying physics mechanism. If confirmed, the observation of J/ψ coherent photoproduction in Pb-Pb collisions at impact parameters smaller than twice the nuclear radius opens new theoretical and experimental challenges and opportunities. In particular, coherent photoproduction accompanying hadronic collisions may provide insight into the dynamics of photoproduction and nuclear reactions, as well as become a novel probe of the Quark-Gluon Plasma.
he first measurement of 3ΛH and 3Λ¯¯¯¯H¯¯¯¯ differential production with respect to transverse momentum and centrality in Pb−Pb collisions at sNN−−−√=5.02~TeV is presented. The 3ΛH has been reconstructed via its two-charged-body decay channel, i.e., 3ΛH→3He+π−. A Blast-Wave model fit of the pT-differential spectra of all nuclear species measured by the ALICE collaboration suggests that the 3ΛH kinetic freeze-out surface is consistent with that of other nuclei. The ratio between the integrated yields of 3ΛH and 3He is compared to predictions from the statistical hadronisation model and the coalescence model, with the latter being favoured by the presented measurements.
In our Galaxy, light antinuclei composed of antiprotons and antineutrons can be produced through high-energy cosmic-ray collisions with the interstellar medium or could also originate from the annihilation of dark-matter particles that have not yet been discovered. On Earth, the only way to produce and study antinuclei with high precision is to create them at high-energy particle accelerators. Although the properties of elementary antiparticles have been studied in detail, the knowledge of the interaction of light antinuclei with matter is limited. We determine the disappearance probability of 3He¯¯¯¯¯¯ when it encounters matter particles and annihilates or disintegrates within the ALICE detector at the Large Hadron Collider. We extract the inelastic interaction cross section, which is then used as input to calculations of the transparency of our Galaxy to the propagation of 3He¯¯¯¯¯¯ stemming from dark-matter annihilation and cosmic-ray interactions within the interstellar medium. For a specific dark-matter profile, we estimate a transparency of about 50%, whereas it varies with increasing 3He¯¯¯¯¯¯ momentum from 25% to 90% for cosmic-ray sources. The results indicate that 3He¯¯¯¯¯¯ nuclei can travel long distances in the Galaxy, and can be used to study cosmic-ray interactions and dark-matter annihilation.
In this letter, measurements of (anti)alpha production in central (0−10%) Pb−Pb collisions at a center-of-mass energy per nucleon−nucleon pair of sNN−−−√ = 5.02 TeV are presented, including the first measurement of an antialpha transverse-momentum spectrum. Owing to its large mass, (anti)alpha production yields and transverse-momentum spectra are of particular interest because they provide a stringent test of particle production models. The averaged antialpha and alpha spectrum is included into a common blast-wave fit with lighter particles, indicating that the (anti)alpha also participates in the collective expansion of the medium created in the collision. A blast-wave fit including only protons, (anti)alpha, and other light nuclei results in a similar flow velocity as the fit that includes all particles. A similar flow velocity, but a significantly larger kinetic freeze-out temperature is obtained when only protons and light nuclei are included in the fit. The coalescence parameter B4 is well described by calculations from a statistical hadronization model but significantly underestimated by calculations assuming nucleus formation via coalescence of nucleons. Similarly, the (anti)alpha-to-proton ratio is well described by the statistical hadronization model. On the other hand, coalescence calculations including approaches with different implementations of the (anti)alpha substructure tend to underestimate the data.
Poster presentation: An important challenge in neuroscience is understanding how networks of neurons go about processing information. Synapses are thought to play an essential role in cellular information processing however quantitative and mathematical models of the underlying physiologic processes that occur at synaptic active zones are lacking. We are generating mathematical models of synaptic vesicle dynamics at a well-characterized model synapse, the Drosophila larval neuromuscular junction. This synapse's simplicity, accessibility to various electrophysiological recording and imaging techniques, and the genetic malleability intrinsic to Drosophila system make it ideal for computational and mathematical studies. We have employed a reductionist approach and started by modeling single presynaptic boutons. Synaptic vesicles can be divided into different pools; however, a quantitative understanding of their dynamics at the Drosophila neuromuscular junction is lacking [4]. We performed biologically realistic simulations of high and low release probability boutons [3] using partial differential equations (PDE) taking into account not only the evolution in time but also the spatial structure in two dimensions (the extension to three dimensions will be implemented soon). PDEs are solved using UG, a program library for the calculation of multi-dimensional PDEs solved using a finite volume approach and implicit time stepping methods leading to extended linear equation systems be solvedwith multi-grid methods [3,4]. Numerical calculations are done on multi-processor computers for fast calculations using different parameters in order to asses the biological feasibility of different models. In preliminary simulations, we modeled vesicle dynamics as a diffusion process describing exocytosis as Neumann streams at synaptic active zones. The initial results obtained with these models are consistent with experimental data. However, this should be regarded as a work in progress. Further refinements will be implemented, including simulations using morphologically realistic geometries which were generated from confocal scans of the neuromuscular junction using NeuRA (a Neuron Reconstruction Algorithm). Other parameters such as glutamate diffusion and reuptake dynamics, as well as postsynaptic receptor kinetics will be incorporated as well.
Mathematical modeling of Arabidopsis thaliana with focus on network decomposition and reduction
(2014)
Systems biology has become an important research field during the last decade. It focusses on the understanding of the systems which emit the measured data. An important part of this research field is the network analysis, investigating biological networks. An essential point of the inspection of these network models is their validation, i.e., the successful comparison of predicted properties to measured data. Here especially Petri nets have shown their usefulness as modeling technique, coming with sound analysis methods and an intuitive representation of biological network data.
A very important tool for network validation is the analysis of the Transition-invariants (TI), which represent possible steady-state pathways, and the investigation of the liveness property. The computational complexity of the determination of both, TI and liveness property, often hamper their investigation.
To investigate this issue, a metabolic network model is created. It describes the core metabolism of Arabidopsis thaliana, and it is solely based on data from the literature. The model is too complex to determine the TI and the liveness property.
Several strategies are followed to enable an analysis and validation of the network. A network decomposition is utilized in two different ways: manually, motivated by idea to preserve the integrity of biological pathways, and automatically, motivated by the idea to minimize the number of crossing edges. As a decomposition may not be preserving important properties like the coveredness, a network reduction approach is suggested, which is mathematically proven to conserve these important properties. To deal with the large amount of data coming from the TI analysis, new organizational structures are proposed. The liveness property is investigated by reducing the complexity of the calculation method and adapting it to biological networks.
The results obtained by these approaches suggest a valid network model. In conclusion, the proposed approaches and strategies can be used in combination to allow the validation and analysis of highly complex biological networks.
The efficient management of large multimedia databases requires the development of new techniques to process, characterize, and search for multimedia objects. Especially in the case of image data, the rapidly growing amount of documents prohibits a manual description of the images’ content. Instead, the automated characterization is highly desirable to support annotation and retrieval of digital images. However, this is a very complex and still unsolved task. To contribute to a solution of this problem, we have developed a mechanism for recognizing objects in images based on the query by example paradigm. Therefore, the most salient image features of an example image representing the searched object are extracted to obtain a scale-invariant object model. The use of this model provides an efficient and robust strategy for recognizing objects in images independently of their size. Further applications of the mechanism are classical recognition tasks such as scene decomposition or object tracking in video sequences.
In den Anwendungsbereichen der Mixed Reality (MR) werden die reale und die virtuelle Welt kombiniert, so dass ein Eindruck der Koexistenz beider Welten entsteht. Meist wird dabei die reale Umgebung durch virtuelle Objekte angereichert, die dem Anwender zusätzliche Informationen bieten sollen. Um die virtuellen Objekte richtig zu positionieren, muss die reale Umgebung erkannt werden. Diese Erkennung der realen Umgebung wird meist durch Bestimmung und Verfolgung von Orientierung und Positionierung der realen Objekte realisiert, was als Tracking bezeichnet wird und einen der wichtigsten Bestandteile für MR-Anwendung darstellt. Ohne die exakte Ausrichtung von realen und virtuellen Objekten, geht die Illusion verloren, dass die virtuellen Objekte Teil der realen Umgebung sind und mit ihr verschmelzen. Markerkombination Das markerbasierte Tracking ist ein Verfahren, das die Bestimmung der Positionierung von realen Objekten durch zusätzliche Markierungen in der realen Umgebung ermöglicht. Diese Markierungen können besonders gut durch Bildanalyseverfahren extrahiert werden und bieten anhand ihrer speziellen Form Positionierungsinformationen. Der Einsatz dieser Trackingtechnologie ist dabei denkbar einfache und kostengünstig. Ein breiter Anwendungsbereich ist durch den kostengünstigen Einsatz dieser Technologien gegeben, allerdings ist das Erstellen von MR-Anwendungen fast ausschließlich MR-Spezialisten vorbehalten, die über Programmierfertigkeiten und spezielle Kenntnisse aus dem MR-Bereich besitzen. Diese Arbeit beschreibt die Entwicklung und Umsetzung der Konzepte, die einem Personenkreis, der lediglich über geringe Kenntnisse von MR-Technologien und deren Anwendung verfügt, den kostengünstigen und einfachen Einsatz von markerbasierten Trackingtechnologien ermöglicht. Die im Rahmen der Arbeit durchgeführte Analyse verweist auf die problematischen Anwendungsfälle des markerbasierten Trackings, die durch die Verdeckung von Markern zustande kommen, in der Beschränkung der Markeranzahl begründet sind, oder durch die Schwankung der Trackingangaben entstehen. Diese Problembereiche sind bei der Entwicklung berücksichtigt worden und können mit Hilfe der entwickelten Konzepte vom Autor bewältigt werden. Das Konzept der Markerkategorien ermöglicht dabei den Einsatz von angepassten Filterungstechniken. Die redundante Markerkombination behebt das Verdeckungsproblem und eliminiert Schwankungen durch das Kombinieren von mehreren Trackinginformationen. Die Gütefunktion ermöglicht die Bewertung von Trackinginformationen und wird zur Gewichtung der Trackingangaben innerhalb einer Markerkombination genutzt. Das Konzept der Markertupel ermöglicht eine Wiederverwendung von Markern, durch den Ansatz der Bereichsunterteilung. Die Konzepte sind in der AMIRE-Umgebung vollständig implementiert und getestet worden. Zum Abschluss ist rückblickend eine kritische Betrachtung der Arbeit, in punkto Vorgehensweise und erreichter Ergebnisse durchgeführt worden.
Dieser Arbeit war zum Ziel gesetzt, Methoden zur Simulation von neuronalen Prozessen zu entwickeln, zu implementieren, einzusetzen und zu vergleichen. Ein besonderes Augenmerk lag dabei auf der Frage, wo eine volle räumliche Auflösung der Modelle benötigt wird und wo darauf zugunsten von vereinfachenden niederdimensionalen Modellen, die wesentlich weniger Ressourcen und mathematischen Sachverstand erfordern, verzichtet werden kann. Außerdem wurde speziell bei der Beschreibung der verschiedenen Modelle für die Elektrik der Nervenzellen das Anliegen verfolgt, deren Zusammenhänge und die Natur vereinfachender Annahmen herauszuarbeiten, um deutlich zu machen, an welchen Stellen Probleme bei der Benutzung der weniger komplexen Modelle auftreten können.
In etlichen Beispielen wurde daraufhin untersucht, inwieweit die Vereinfachung auf ein eindimensionales Kabelmodell sowie der Verzicht auf die Betrachtung einzelner Ionensorten die realistische Darstellung der zellulären Elektrik beeinträchtigen können. Dabei stellte sich heraus, dass alle betrachteten Modelle für das rein elektrische Verhalten der Neuronen im Wesentlichen dieselben Ergebnisse liefern, weshalb zu dessen Simulation in den allermeisten Fällen ein 1D-Kabelmodell völlig ausreichend und angezeigt sein dürfte.
Nur wenn Größen von Interesse sind, die in diesem Modell nicht erfasst werden, etwa das Außenraumpotential oder die Ionenkonzentrationen, muss auf genauere Modelle zurückgegriffen werden. Außerdem ist in einer Konvergenzstudie exemplarisch vorgeführt worden, dass bereits eine recht grobe Darstellung der zugrundeliegenden Rechengitter genügt, um korrekte Ergebnisse bei der Simulation der rein elektrischen Signale sicherzustellen.
In scharfem Kontrast steht hierzu die Simulation von einzelnen Ionen-Dynamiken. Bereits in der Untersuchung des Poisson-Nernst-Planck-Modells für das Membranpotential erwies sich, dass für eine korrekte Simulation der diffusiven Anteile der Ionenbewegung wesentlich feinere Gitter benötigt werden.
Noch viel deutlicher wurde dies in Simulationen von Calcium-Wellen in Dendriten, wo -- neben anderen Einsichten -- aufgezeigt werden konnte, dass nicht nur eine feine axiale
(und Zeit-) Auflösung der Dendritengeometrie zur Sicherstellung exakter Ergebnisse notwendig ist, sondern auch die räumliche Auflösung in die übrigen Dimensionen wichtig ist, weswegen eine eindimensionale Kabeldarstellung der Calcium-Dynamik erheblich fehlerbehaftet und
(jedenfalls im Zusammenhang mit Ryanodin-Rezeptorkanälen) von deren Nutzung dringend abzuraten ist. Auch die Darstellung von Kanälen als eine kontinuierliche Dichte in der Membran kann, wie darüber hinaus vorgeführt wurde, problematisch sein.
Ihre exaktere Modellierung, etwa durch Einbettung auch probabilistischer Einzelkanaldarstellungen in das räumliche Modell sollte in zukünftigen Arbeiten noch mehr thematisiert werden.
Mit Blick auf die Wiederverwendbarkeit bereits implementierter Funktionalität innerhalb dieser Arbeiten wurden spezielle Teile dieser Funktionalität hier in einem gesonderten
Kapitel genauer beschrieben. Als komplexes Beispiel für das, was simulationstechnisch bereits im Bereich des Machbaren
liegt, und gleichsam für eine Anwendung, die zeigt, wie möglichst viele der im Rahmen dieser Arbeit entwickelten Methoden miteinander kombiniert werden können, wurde die
Calcium-Dynamik eines kompletten Dendriten innerhalb eines großen aktiven neuronalen Netzwerks simuliert.
Background: Signal transduction pathways are important cellular processes to maintain the cell’s integrity. Their imbalance can cause severe pathologies. As signal transduction pathways feature complex regulations, they form intertwined networks. Mathematical models aim to capture their regulatory logic and allow an unbiased analysis of robustness and vulnerability of the signaling network. Pathway detection is yet a challenge for the analysis of signaling networks in the field of systems biology. A rigorous mathematical formalism is lacking to identify all possible signal flows in a network model.
Results: In this paper, we introduce the concept of Manatee invariants for the analysis of signal transduction networks. We present an algorithm for the characterization of the combinatorial diversity of signal flows, e.g., from signal reception to cellular response. We demonstrate the concept for a small model of the TNFR1-mediated NF- κB signaling pathway. Manatee invariants reveal all possible signal flows in the network. Further, we show the application of Manatee invariants for in silico knockout experiments. Here, we illustrate the biological relevance of the concept.
Conclusions: The proposed mathematical framework reveals the entire variety of signal flows in models of signaling systems, including cyclic regulations. Thereby, Manatee invariants allow for the analysis of robustness and vulnerability of signaling networks. The application to further analyses such as for in silico knockout was shown. The new framework of Manatee invariants contributes to an advanced examination of signaling systems.
We determine the magnetic susceptibility of thermal QCD matter by means of first principles lattice simulations using staggered quarks with physical masses. A novel method is employed that only requires simulations at zero background field, thereby circumventing problems related to magnetic flux quantization. After a careful continuum limit extrapolation, diamagnetic behavior (negative susceptibility) is found at low temperatures and strong paramagnetism (positive susceptibility) at high temperatures. We revisit the decomposition of the magnetic susceptibility into spin- and orbital angular momentum- related contributions. The spin term — related to the normalization of the photon lightcone distribution amplitude at zero temperature — is calculated non-perturbatively and extrapolated to the continuum limit. Having access to both the full magnetic susceptibility and the spin term, we calculate the orbital angular momentum contribution for the first time. The results reveal the opposite of what might be expected based on a free fermion picture. We provide a simple parametrization of the temperature- and magnetic field-dependence of the QCD equation of state that can be used in phenomenological studies.
We investigate the QCD phase diagram for nonzero background magnetic fields using first-principles lattice simulations. At the physical point (in terms of quark masses), the thermodynamics of this system is controlled by two opposing effects: magnetic catalysis (enhancement of the quark condensate) at low temperature and inverse magnetic catalysis (reduction of the condensate) in the transition region. While the former is known to be robust and independent of the details of the interactions, inverse catalysis arises as a result of a delicate competition, effective only for light quarks. By performing simulations at different quark masses, we determine the pion mass above which inverse catalysis does not take place in the transition region anymore. Even for pions heavier than this limiting value — where the quark condensate undergoes magnetic catalysis — our results are consistent with the notion that the transition temperature is reduced by the magnetic field. These findings will be useful to guide low-energy models and effective theories of QCD.
In the last years, much effort went into the design of robust anaphor resolution algorithms. Many algorithms are based on antecedent filtering and preference strategies that are manually designed. Along a different line of research, corpus-based approaches have been investigated that employ machine-learning techniques for deriving strategies automatically. Since the knowledge-engineering effort for designing and optimizing the strategies is reduced, the latter approaches are considered particularly attractive. Since, however, the hand-coding of robust antecedent filtering strategies such as syntactic disjoint reference and agreement in person, number, and gender constitutes a once-for-all effort, the question arises whether at all they should be derived automatically. In this paper, it is investigated what might be gained by combining the best of two worlds: designing the universally valid antecedent filtering strategies manually, in a once-for-all fashion, and deriving the (potentially genre-specific) antecedent selection strategies automatically by applying machine-learning techniques. An anaphor resolution system ROSANA-ML, which follows this paradigm, is designed and implemented. Through a series of formal evaluations, it is shown that, while exhibiting additional advantages, ROSANAML reaches a performance level that compares with the performance of its manually designed ancestor ROSANA.
LSTMVoter : chemical named entity recognition using a conglomerate of sequence labeling tools
(2019)
Background: Chemical and biomedical named entity recognition (NER) is an essential preprocessing task in natural language processing. The identification and extraction of named entities from scientific articles is also attracting increasing interest in many scientific disciplines. Locating chemical named entities in the literature is an essential step in chemical text mining pipelines for identifying chemical mentions, their properties, and relations as discussed in the literature. In this work, we describe an approach to the BioCreative V.5 challenge regarding the recognition and classification of chemical named entities. For this purpose, we transform the task of NER into a sequence labeling problem. We present a series of sequence labeling systems that we used, adapted and optimized in our experiments for solving this task. To this end, we experiment with hyperparameter optimization. Finally, we present LSTMVoter, a two-stage application of recurrent neural networks that integrates the optimized sequence labelers from our study into a single ensemble classifier.
Results: We introduce LSTMVoter, a bidirectional long short-term memory (LSTM) tagger that utilizes a conditional random field layer in conjunction with attention-based feature modeling. Our approach explores information about features that is modeled by means of an attention mechanism. LSTMVoter outperforms each extractor integrated by it in a series of experiments. On the BioCreative IV chemical compound and drug name recognition (CHEMDNER) corpus, LSTMVoter achieves an F1-score of 90.04%; on the BioCreative V.5 chemical entity mention in patents corpus, it achieves an F1-score of 89.01%.
Availability and implementation: Data and code are available at https://github.com/texttechnologylab/LSTMVoter.