Search

Aspect-based sentiment analysis (ABSA) is a fine-grained sentiment analysis task designed to predict the sentiment polarity of each aspect term in a text. Recent research mainly uses neural networks to model text and utilizes attention mechanisms to interact for associate aspect terms and context to obtain more effective feature representation. However, the general attention mechanism is easy to lose the original information. Besides, in the multi-aspect text, the sentiment information of other aspect terms interferes with the sentiment analysis of the current aspect term likely. In this paper, we propose two models named RA-CNN and RAO-CNN for ABSA tasks. In RA-CNN, we apply CNN to model the aspect term and utilize a specially designed residual attention mechanism to interact with the text. Based on the RA-CNN, RAO-CNN adds other aspect terms module, which can reduce interference of sentiment information related to other aspect terms in the multi-aspect text. To verify the proposed models’ effectiveness, we conduct a large number of experiments and comparisons on seven public datasets. Experimental results show that our proposed models are useful and achieve state-of-the-art results.

Aspect-based sentiment analysis can predict the sentiment polarity of specific aspect terms in the text. Compared to general sentiment analysis, it extracts more useful information and analyzes the sentiment more accurately in the comment text. Many previous approaches use long short-term memory networks with attention mechanisms to directly learn aspect-specific representations and model comment text. However, these methods always ignore the importance of the aspect terms position and interactive information between the aspect terms and other words. To address these issues, we propose an improved model based on convolutional neural networks. First, a novel relative position encode layer can integrate the relative position information of specific aspect terms validly in a text. Second, by using the aspect attention mechanism, the semantic relationship between aspect terms and words in the text is fully considered. To verify the effectiveness of the proposed models, we conduct a large number of experiments and comparisons on seven public datasets. The experimental results show that this model outperforms to other state-of-the-art methods.

In this paper, a new generalized step-up multilevel DC-AC converter is proposed, which is suitable for applications with low-voltage input sources such as photovoltaic power generation and electric vehicles. This inverter can achieve a high voltage gain by controlling the series-parallel conversion of DC power supply and capacitors. Only one DC voltage source and a few power devices are employed. The maximum output voltage and the number of output levels can be further increased through the switched-capacitor unit's extension and the submodule cascaded extension. Moreover, the capacitor voltages are self-balanced without complicated voltage control circuits. The complementary operating mechanism between each pair of switches simplifies the modulation algorithm. The inductive-load ability is fully taken into account in the proposed inverter. Additionally, a remarkable characteristic of the inverter is that the charging and discharging states among different capacitors are synchronous, which reduces the voltage ripple of the frontend capacitors. The circuit structure, the working principle, the modulation strategy, the capacitors and losses analysis are presented in detail. Afterwards, the advantages of the proposed inverter are analyzed by comparing with other recently proposed inverters. Finally, the steady-state and dynamic performance of the proposed inverter is verified and validated by simulation and experiment. 1

Because social networks have become a vital part of people’s lives, cyberbullying becomes the most common risk encountered by young people on social networking platforms and raised serious concerns in society. Over the past few decades, most existing work on cyberbullying has focused on text analysis. Yet, the cyberbullying develops into multi-objective, multi-channel, and multi-form. Traditional text analysis methods cannot satisfy the diversity of bullying data in social networks. To deal with the new type of cyberbullying, we propose a multi-modal detection framework that takes into multi-modal information(e.g., image, video, comments, time) on social networks. Specifically, we not only extract textual features but also use the hierarchical attention networks to capture the session feature in social networks and encode several media information(e.g., video, image). Based on these features, we model the multi-modal cyberbullying detection framework to solve the new form of cyberbullying. Experimental analysis on two real-world datasets shows that our framework outperforms several existing state-of-the-art models.

A large area nanopore array with controllable consistency has been fabricated by direct laser writing (DLW) technique with ultralow one-photon absorption (LOPA). In this technique, a doughnut-shape beam generated by 532 nm continuous-wave laser through a vortex phase plate was focused into the thin film of SU8 photoresist by a high numerical aperture objective. The low absorption of the photoresist at the excitation wavelength allowed controllable fabricating structures with sub-diffraction-limit feature size and high aspect ratio. With point-by-point scanning fabrication, nanopore array with the pore's internal diameter, about ten percent of incident laser wavelength, far smaller than Abbe's diffraction limit was achieved. The influences of the exposure intensity and exposure time on the fabricated nanopore size and shape were also investigated by scanning electron microscope (SEM).

This paper studies the properties of a kind of portable ultra-bright microfocus x-ray source with the Monte-Carlo method in detail. The new x-ray source consists of an electron-emission system, an electrostatic focusing system and a metal target. A crystal Lanthanum Hexaboride cathode, a Wehnelt grid and an extracted electrode compose the triode electrode electron-gun system. Two equal radius cylinder electrodes form the focusing system. The key factors determining the focus properties of the electron beam such as the ratio Dw/H, grid bias Vg, and the properties of the extracted electrode are numerically studied. The calculated results reveal that when Dw/H, Vg, the length of the extracted electrode, and the distance between the grid and the extracted electrode equals 5, –0.6 kV, 10 mm, and 8 mm respectively, the electron beam focal spot can be concentrated down to 9 μm in radius and a reasonable focal length about 72.5 mm can be achieved, at the same time, the cathode emission currents can be as high as 30 mA.

This paper reports that when an intense extraordinary-polarized laser beam illuminates a photorefractive BaTiO3 crystal, the dynamic beam fanning light is formed to be a thermal-like light source with a long correlation time and wide spectral bandwidth. The experimental results of the first- and second-order double-slit interference with such photorefractive fanning light source, can be understood with the theoretical simulation in terms of Hanbury-Brown and Twiss effect.

We propose a special two-photon state which is completely transparent in a 50/50 beam splitter. This effect is caused by the destructive two-photon interference and shows the signature of photon entanglement. We find that the symmetry of the two-photon spectrum plays the key role for the properties of two-photon interference.

The sub-diffraction-limit spatially structured light patterns have attracted more and more attention for their important applications in many frontier scientific fields. The present paper aims at developing sub-diffraction-limit spatially structured beam patterns which might have great potential to improve the light performance in fields such as super resolution imagery, optical tweezer, micro/nano lithography, etc. Here, a variety of spatially structured beam patterns are obtained by the phase modulation of polarized beams and studied in detail experimentally and numerically. Firstly, a new kind of phase plate, which combines the merits of circular and vortex 2 phase plates, is proposed based on the wave front design; it is composed of two spiral-shaped phase plates with their phases changing from 0 to 2 and - to , respectively. Later, the phase plate is applied to the circularly polarized Gaussian beam modulation in a high NA system. By combining a self-made circular with a commercial vortex 2 phase plate, the designed new phase plate is implemented in the experiment. The morphology of the spatially structured light pattern, which is generated on the focal plane, is observed by a CCD camera in the experiment. The beam pattern presents a donut shape on the focal plane, while the dimension of the donut-shaped pattern becomes smaller as the imaging plane axially deviates from the focal plane. It is found that the beam patterns captured in experiment highly consist with the numerical simulation results carried out by the vectorial diffraction integral theory. It can be deduced that the spatially structured beam is capillary-shaped. Meanwhile, at the two ends of the capillary-shaped beam, the inner diameter is smaller than the diffraction limitation. Furthermore, the structured beam pattern presents a spatial voxel distribution with center and axis symmetry. Finally, the characteristics of the spatially structured beam patterns, which are generated by modulating circular, linear, radial and azimuthal polarized beams with the new designed phase plate, are analyzed and discussed in detail. It is found that for circular, linear, radial and azimuthal polarization, the full widths at half maximum (FWHMs) of the minimum dark spots in the horizontal direction are 0.31, 0.32, 0.24 and 0.36, respectively. On the optical axis, the FWHMs of the dark spots created by linearly, radially and azimuthally polarized light, are 0.8, 0.78 and 0.76 , respectively, and no axial intensity is found with circularly polarized beam incidence.

We propose a novel wide-field fluorescence sectioning microscope based on polarization filtering image enhancement and dynamic speckle illumination in this article. A new type of polarization filtering image enhancement technology, based on the differences in polarization characteristic between excitation light and fluorescence, is used to filter out the excitation light and enhance image quality. The dynamic speckle illumination is employed to achieve the wide field fluorescence sectioning which has many advantages, such as simple configuration, low cost, rapid response and easy operation. The experimental results indicate that our filtering method can be utilized to improve the image quality significantly, and the dynamic using speckle illumination can be employed to achieve optical sectioning with a wide field and a high vertical resolution capability. Our study is not only to enrich the technological method to extract the weak fluorescence signal from the strong excitation light, but also to provide a significant reference for developing a light wavelength tunable multi-spectral fluorescence microscope with rapid response.

We show that the spontaneous formation of transverse spatial patterns due to diffraction in lasers can be described by a phase diffusion equation of the Kuramoto-Sivashinsky kind. Hence the origin of the pattern formation lies in spatial instabilities of the phase of the complex slowly varying laser field amplitude.