Open Information Extraction (OIE) is the task of extracting facts from sentences in the form of relations and their corresponding arguments in schema-free manner. Intrinsic performance of OIE systems is difficult to measure due to the incompleteness
of existing OIE benchmarks: the ground truth extractions do not group all acceptable surface realizations of the same fact that can be extracted from a sentence. To measure performance of OIE systems more realistically, it is necessary to manually annotate complete facts (i.e., clusters of all acceptable surface realizations of the same fact) from input sentences. We propose AnnIE: an interactive annotation platform that facilitates such challenging annotation tasks and supports creation of complete fact-oriented OIE evaluation benchmarks. AnnIE is modular and flexible in order to support different use case scenarios (i.e., benchmarks covering different types of facts). We use AnnIE to build two complete OIE benchmarks: one with verb-mediated facts and another with facts encompassing named entities. Finally, we evaluate several OIE systems on our complete benchmarks created with AnnIE. Our results suggest that existing incomplete benchmarks are overly lenient, and that OIE systems are not as robust as previously reported. We publicly release AnnIE under non-restrictive license.
Intrinsic evaluations of OIE systems are carried out either manually -- with human evaluators judging the correctness of extractions -- or automatically, on standardized benchmarks. The latter, while much more cost-effective, is less reliable, primar
ily because of the incompleteness of the existing OIE benchmarks: the ground truth extractions do not include all acceptable variants of the same fact, leading to unreliable assessment of models performance. Moreover, the existing OIE benchmarks are available for English only. In this work, we introduce BenchIE: a benchmark and evaluation framework for comprehensive evaluation of OIE systems for English, Chinese and German. In contrast to existing OIE benchmarks, BenchIE takes into account informational equivalence of extractions: our gold standard consists of fact synsets, clusters in which we exhaustively list all surface forms of the same fact. We benchmark several state-of-the-art OIE systems using BenchIE and demonstrate that these systems are significantly less effective than indicated by existing OIE benchmarks. We make BenchIE (data and evaluation code) publicly available.
Synthetic dimensions in photonic structures provide unique opportunities for actively manipulating light in multiple degrees of freedom. Here, we theoretically explore a dispersive waveguide under the dynamic phase modulation that supports single pul
se manipulations in the synthetic (2+1) dimensions. Compared with the counterpart of the conventional (2+1) space-time, we explore temporal diffraction and frequency conversion in a synthetic time-frequency space while the pulse evolves along the spatial dimension. By introducing the effective gauge potential well for photons in the synthetic time-frequency space with the control of the modulation phase, we show that a rich set of pulse propagation behaviors can be achieved, including confined pulse propagation, fast/slow light, and pulse compression. With the additional nonlinear oscillation subject to the effective force along the frequency axis of light, we provide an exotic approach for actively manipulating the single pulse in both temporal and spectral domains, which shows the great promise for applications of the pulse processing and optical communications in integrated photonics.
Let $mathscr{G}_{n,beta}$ be the set of graphs of order $n$ with given matching number $beta$. Let $D(G)$ be the diagonal matrix of the degrees of the graph $G$ and $A(G)$ be the adjacency matrix of the graph $G$. The largest eigenvalue of the nonneg
ative matrix $A_{alpha}(G)=alpha D(G)+A(G)$ is called the $alpha$-spectral radius of $G$. The graphs with maximal $alpha$-spectral radius in $mathscr{G}_{n,beta}$ are completely characterized in this paper. In this way we provide a general framework to attack the problem of extremal spectral radius in $mathscr{G}_{n,beta}$. More precisely, we generalize the known results on the maximal adjacency spectral radius in $mathscr{G}_{n,beta}$ and the signless Laplacian spectral radius.
Let $G$ be a graph. For a subset $X$ of $V(G)$, the switching $sigma$ of $G$ is the signed graph $G^{sigma}$ obtained from $G$ by reversing the signs of all edges between $X$ and $V(G)setminus X$. Let $A(G^{sigma})$ be the adjacency matrix of $G^{sig
ma}$. An eigenvalue of $A(G^{sigma})$ is called a main eigenvalue if it has an eigenvector the sum of whose entries is not equal to zero. Let $S_{n,k}$ be the graph obtained from the complete graph $K_{n-r}$ by attaching $r$ pendent edges at some vertex of $K_{n-r}$. In this paper we prove that there exists a switching $sigma$ such that all eigenvalues of $G^{sigma}$ are main when $G$ is a complete multipartite graph, or $G$ is a harmonic tree, or $G$ is $S_{n,k}$. These results partly confirm a conjecture of Akbari et al.
Automatic International Classification of Diseases (ICD) coding is defined as a kind of text multi-label classification problem, which is difficult because the number of labels is very large and the distribution of labels is unbalanced. The label-wis
e attention mechanism is widely used in automatic ICD coding because it can assign weights to every word in full Electronic Medical Records (EMR) for different ICD codes. However, the label-wise attention mechanism is computational redundant and costly. In this paper, we propose a pseudo label-wise attention mechanism to tackle the problem. Instead of computing different attention modes for different ICD codes, the pseudo label-wise attention mechanism automatically merges similar ICD codes and computes only one attention mode for the similar ICD codes, which greatly compresses the number of attention modes and improves the predicted accuracy. In addition, we apply a more convenient and effective way to obtain the ICD vectors, and thus our model can predict new ICD codes by calculating the similarities between EMR vectors and ICD vectors. Extensive experiments show the superior performance of our model. On the public MIMIC-III dataset and private Xiangya dataset, our model achieves micro f1 of 0.583 and 0.806, respectively, which outperforms other competing models. Furthermore, we verify the ability of our model in predicting new ICD codes. The case study shows how pseudo label-wise attention works, and demonstrates the effectiveness of pseudo label-wise attention mechanism.
The fractional matching number of a graph G, is the maximum size of a fractional matching of G. The following sharp lower bounds for a graph G of order n are proved, and all extremal graphs are characterized in this paper. (1)The sum of the fractiona
l matching number of a graph G and the fractional matching number of its complement is not less than n/2 , where n is not less than 2. (2) If G and its complement are non-empty, then the sum of the fractional matching number of a graph G and the fractional matching number of its complement is not less than (n+1)/2, where n is not less than 28. (3) If G and its complement have no isolated vertices, then the sum of the fractional matching number of a graph G and the fractional matching number of its complement is not less than (n+4)/2, where n is not less than 28.
Metasurfaces are planar structures that can manipulate the amplitude, phase and polarization (APP) of light at subwavelength scale. Although various functionalities have been proposed based on metasurface, a most general optical control, i.e., indepe
ndent complex-amplitude (amplitude and phase) control of arbitrary two orthogonal states of polarizations, has not yet been realized. Such level of optical control can not only cover the various functionalities realized previously, but also enable new functionalities that are not feasible before. Here, we propose a single-layer dielectric metasurface to realize this goal and experimentally demonstrate several advanced functionalities, such as two independent full-color printing images under arbitrary elliptically orthogonal polarizations and dual sets of printing-hologram integrations. Our work opens the way for a wide range of applications in advanced image display, information encoding, and polarization optics.
A dynamically-modulated ring system with frequency as a synthetic dimension has been shown to be a powerful platform to do quantum simulation and explore novel optical phenomena. Here we propose synthetic honeycomb lattice in a one-dimensional ring a
rray under dynamic modulations, with the extra dimension being the frequency of light. Such system is highly re-configurable with modulation. Various physical phenomena associated with graphene including Klein tunneling, valley-dependent edge states, effective magnetic field, as well as valley-dependent Lorentz force can be simulated in this lattice, which exhibits important potentials for manipulating photons in different ways. Our work unveils a new platform for constructing the honeycomb lattice in a synthetic space, which holds complex functionalities and could be important for optical signal processing as well as quantum computing.
The notion of topological phases extended to dynamical systems stimulates extensive studies, of which the characterization of non-equilibrium topological invariants is a central issue and usually necessitates the information of quantum dynamics in bo
th the time and spatial dimensions. Here we combine the recently developed concepts of the dynamical classification of topological phases and synthetic dimension, and propose to efficiently characterize photonic topological phases via holographic quench dynamics. A pseudo spin model is constructed with ring resonators in a synthetic lattice formed by frequencies of light, and the quench dynamics is induced by initializing a trivial state which evolves under a topological Hamiltonian. Our key prediction is that the complete topological information of the Hamiltonian is extracted from quench dynamics solely in the time domain, manifesting holographic features of the dynamics. In particular, two fundamental time scales emerge in the quench dynamics, with one mimicking the Bloch momenta of the topological band and the other characterizing the residue time evolution of the state after quench. For this a dynamical bulk-surface correspondence is obtained in time dimension and characterizes the topology of the spin model. This work also shows that the photonic synthetic frequency dimension provides an efficient and powerful way to explore the topological non-equilibrium dynamics.
