Search

With the rapid development of e-commerce, online consumption has become a mainstream form of consumption in recent years. Text sentiment analysis for a large number of customer reviews on the e-commerce platform can dramatically improve the customers' consumption experience. Although the sentiment analysis approaches based on deep neural network can achieve higher accuracy without human-design features compared with traditional sentiment analysis methods, the accuracy still cannot meet the demand and the training suffers from the issues of over-fitting, vanishing gradient, etc. In this paper, a novel sentiment analysis model named MBGCV is designed to alleviate these problems and improve the accuracy, MBGCV employs a multichannel paradigm and integrates Bidirectional Gated Recurrent Unit (BiGRU), Convolutional Neural Network (CNN) and Variational Information Bottleneck (VIB). The multichannel is exploited to extract multi-grained sentiment features. In each channel, BiGRU is utilized to extract context information, and then CNN is adopted to extract local features. Furthermore, the model combines the advantages of VIB and Maxout activation function to overcome shortcomings such as over-fitting, vanishing gradient in existing sentiment analysis models. By using real review datasets, we carry out extensive experiments to demonstrate that our proposed model can achieve superior accuracy and improve the performance of text sentiment analysis.

Thin, lightweight, and flexible textile pressure sensors with the ability to precisely detect the full range of faint pressure (via a facile electrophoretic deposition (EPD) approach. High-density FCNT is evenly wrapped and chemically bonded to the fiber surface during the EPD process, forming a conductive hierarchical fiber/FCNT matrix. Benefiting from the abundant yet firm contacting points, point-to-point contacting mode, and high elastic modulus of both PI and CNT, the proposed PI/FCNT pressure sensor exhibits ultra-high sensitivity (3.57 MPa− 1), ultra-wide linearity (3.24 MPa), exceptionally broad sensing range (~ 45 MPa), and long-term stability (> 4000 cycles). Furthermore, under a high working temperature of 200 ºC, the proposed sensor device still shows an ultra-high sensitivity of 2.64 MPa− 1 within a wide linear range of 7.2 MPa, attributing to its intrinsic high-temperature-resistant properties of PI and CNT. Thanks to these merits, the proposed PI/FCNT(EPD) pressure sensor could serve as an E-skin device to monitor the human physiological information, precisely detect tiny and extremely high pressure, and can be integrated into an intelligent mechanical hand to detect the contact force under high-temperature (> 300 ºC), endowing it with high applicability in the fields of real-time health monitoring, intelligent robots, and harsh environments.

Adopting semiconductor materials with light-trapping architectures as the surface-enhanced Raman spectroscopy (SERS)-active substrates has received increasing attention, in which the multiple reflection and scattering of electromagnetic waves can improve the enhancement factor. However, the fabrication of these semiconductor SERS-active substrates commonly requires complicated processes, and these methods normally result in non-uniform and isolated particles/flakes, which have fundamental difficulties in meeting the practical needs in high performance and reliable SERS substrates. Herein, we demonstrate that SnSe2 nanoplate arrays (NPAs) via self-assembled growth can serve as uniform, highly sensitive and reliable SERS substrates. The cavity formed by the SnSe2 NPAs can trap light efficiently (∼96%) and thus improve the enhancement factor. Benefiting from the synergistic effect of the charge-transfer process and enhanced light trapping, the resulting SERS substrates based on pure SnSe2 NPAs show an ultralow detection limit (1 × 10−12 M), a high enhancement factor (6.33 × 106) and excellent uniformity (relative standard deviations down to 7.7%), demonstrating one of the highest sensitivities amongst the reported semiconductor SERS substrates. Furthermore, the effects of different SnSe2 structural configurations (planar vs. cavity), the height and tilt angle of the SnSe2 NPAs on the performance of SERS detection are systematically investigated. It is discovered that the SERS performance is strongly dependent on both the light-trapping ability and the absorption loss. We believe that our results not only provide an effective strategy to obtain tunable, uniform and high performance SERS substrates, but also lead to further understanding of designing 3D light trapping architectures.

Objective To implant 80 mm-long artificial esophagi constructed of biomaterial in dogs, observe the perioperative survival rates and the incidence of postoperative complications, and study the mechanisms of postoperative healing. Methods Specimens of the implanted esophagus, the “neo-esophagi”, were taken for histopathologic study 1, 3, 6, 12, 24 months after operation. Results The incidence of anastomotic leakage after the artificial esophagus implantation was 3.33%. The perioperative survival rate was 96.67%. The incidence of postoperative stenosis in the “neo-esophagi” was 81.48%; the stenoses were treated by expanding with esophagoscopy and implanting a stent. Epithelization of the mucosa in the “neo-esophagi” was completed in 3 to 6 months after surgery. Structures such as submucosal muscle layers, mucous glands, nerve fibers, capillaries, etc . were regenerated after 12 months, and then reconstruction of the fibrous connective tissue layer was completed. Conclusion Implanting a biomaterial artificial esophagus accomplishes safe reconstruction of defects in the esophagus. Advanced cellular structure of “neo-esophagus” can be regenerated after 1 year. Postoperative stenosis, which is related to hyperplasia and retraction of scar tissue, is still the most common complications which limiting the clinical application of the artificial esophagus.

There is a surge in the use of the solar chimney power plant in the recent years which accomplishes the task of converting solar energy into kinetic energy. As the existing models are insufficient to accurately describe the mechanism, a more comprehensive model is advanced in this paper to evaluate the performance of a solar chimney power plant system, in which the effects of various parameters on the relative static pressure, driving force, power output and efficiency have been further investigated. Using the solar chimney prototype in Manzanares, Spain, as a practical example, the numerical studies are performed to explore the geometric modifications on the system performance, which show reasonable agreement with the analytical model. Copyright © 2005 John Wiley & Sons, Ltd.

This paper designs and implies a high precision FBG demodulation system which based on F-P etalon. In order to reduce the influence of the temperature drift effect, the peristaltic effect, and the nonlinear effect of F-P filter in traditional tunable filter method, F-P etalon is added as dynamical calibration and wavelength reference. Meanwhile segmentation demodulation which uses ASE spectral characteristics is applied to achieve high accuracy of the center wavelength of FBG. The experiment shows that the stability, resolution are 0.65pm, 0.23pm, respectively. Key words: fiber optics; fiber Bragg grating sensor system; tunable Fabry-Perot filter; F-P etalon; spectrum segmentation demodulation

The red phosphors play a vital role in controlling luminous efficiency, color rendering index and adjustability in color temperature of white light emitting diodes (w-LEDs). Here, we demonstrate a type of red-emitting phosphors Eu(3+)-doped barium molybdate phosphors effectively excited by the blue light (464nm). These phosphors present the strong emission spectra around 613nm and their color coordinate values are close to the NTSC standard values. More importantly, by constructing a complete 3003×3003 energy matrix by an effective operator Hamiltonian including the free ion and crystal field interactions, for the first time, the fluorescent mechanism for Eu(3+) ions at the tetragonal (S4) Ba(2+) site of BaMoO(4) crystal were investigated by a complete diagonalization (of energy matrix) method (CMD). The fitting values are close to the experimental results, demonstrating the feasibility of using the complete diagonalization method to study the phosphors for w-LEDs.

The weakening groove becomes the weaker link in separating unit since it bears axial load, bending moment and shearing force during operating. The strain distribution of weakening groove is obtained by ARAMIS system based on 3D digital speckle correlation method (DSCM). The result indicates that there is stress concentrate band in each groove, and the strain increased from the lower one to the upper ones. The result also demonstrates that the ARAMIS system performed with good results, especially in 3D deformation and stress concentrate measurement.

In free space optical communication (FSO) system, the acquisition, tracking and pointing (ATP) is very important for system performance. In AirLink-E100 system, we use position sensitive detector (PSD), which has good position linearity and big detection area, to measure angle deviation in ATP. But due to the background light and the dark current, the angle precision of PSD direct detection can not match the requirement of the precision angle measurement. In this article, difference methods achieved by analog signal processing and digital signal processing are adopted to solve this problem and improve the performance of ATP. By experiment, the validity of this method has been verified.