One key task of fine-grained sentiment analysis of product reviews is to extract product aspects or features that users have expressed opinions on. This paper focuses on supervised aspect extraction using deep learning. Unlike other highly sophisticated supervised deep learning models, this paper proposes a novel and yet simple CNN model 1 employing two types of pre-trained embeddings for aspect extraction: general-purpose embeddings and domain-specific embeddings. Without using any additional supervision, this model achieves surprisingly good results, outperforming state-of-the-art sophisticated existing methods. To our knowledge, this paper is the first to report such double embeddings based CNN model for aspect extraction and achieve very good results.
A highly stable porous lanthanide metal-organic framework, Y(BTC)(H2O).4.3H2O (BTC = 1,3,5-benzenetricarboxylate), with pore size of 5.8 A has been constructed and investigated for hydrogen storage. Gas sorption measurements show that this porous MOF exhibits highly selective sorption behaviors of hydrogen over nitrogen gas molecules and can take up hydrogen of about 2.1 wt % at 77 K and 10 bar. Difference Fourier analysis of neutron powder diffraction data revealed four distinct D2 sites that are progressively filled within the nanoporous framework. Interestingly, the strongest adsorption sites identified are associated with the aromatic organic linkers rather than the open metal sites, as occurred in previously reported MOFs. Our results provide for the first time direct structural evidence demonstrating that optimal pore size (around 6 A, twice the kinetic diameter of hydrogen) strengthens the interactions between H2 molecules and pore walls and increases the heat of adsorption, which thus allows for enhancing hydrogen adsorption from the interaction between hydrogen molecules with the pore walls rather than with the normally stronger adsorption sites (the open metal sites) within the framework. At high concentration H2 loadings (5.5 H2 molecules (3.7 wt %) per Y(BTC) formula), H2 molecules form highly symmetric novel nanoclusters with relatively short H2-H2 distances compared to solid H2. These observations are important and hold the key to optimizing this new class of rare metal-organic framework (RMOF) materials for practical hydrogen storage applications.
Nitrogen-rich transition-metal nitrides hold great promise to be the next-generation catalysts for clean and renewable energy applications. However, incorporation of nitrogen into the crystalline lattices of transition metals is thermodynamically unfavorable at atmospheric pressure; most of the known transition metal nitrides are nitrogen-deficient with molar ratios of N:metal less than a unity. In this work, we have formulated a high-pressure route for the synthesis of a nitrogen-rich molybdenum nitride through a solid-state ion-exchange reaction. The newly discovered nitride, 3R-MoN2, adopts a rhombohedral R3m structure, isotypic with MoS2. This new nitride exhibits catalytic activities that are three times more active than the traditional catalyst MoS2 for the hydrodesulfurization of dibenzothiophene and more than twice as high in the selectivity to hydrogenation. The nitride is also catalytically active in sour methanation of syngas with >80% CO and H2 conversion at 723 K. Our formulated route for the synthesis of 3R-MoN2 is at a moderate pressure of 3.5 GPa and, thus, is feasible for industrial-scale catalyst production.
Self-organization of colloidal Pt nanocubes into two types of distinct ordered superlattices, simple-cubic and body-centered-tetragonal structures, has been achieved using a home-built setup. Detailed translational and orientational characteristics of these superstructures were determined using a transmission electron microscopy tomographic technique with 3D reconstruction analysis. The formation of these distinct superlattices is the result of a delicate choice of solvent (i.e., aliphatic hexane or aromatic toluene hydrocarbons), which serves as a dispersion medium to fine-tune the relative strengths of ligand-ligand and ligand-solvent interactions during the self-assembly process. This work provides important insights into the effects of ligand-solvent interactions on superlattice formation from nonspherical nanoparticles.
The structures of ordered and disordered -eucryptite have been determined from Rietveld analysis of powder synchrotron x-ray and neutron diffraction data over a temperature range of 20 to 873 K. On heating, both materials show an expansion within the (001) plane and a contraction along the c axis. However, the anisotropic character of the thermal behavior of ordered -eucryptite is much more pronounced than that of the disordered compound; the linear expansion coefficients of the ordered and disordered phases are ␣ a ס 7.26 × 10 −6 K −1 ; ␣ c ס −16.35 × 10 −6 K −1 , and ␣ a ס 5.98 × 10 −6 K −1 ; ␣ c ס −3.82 × 10 −6 K −1 , respectively. The thermal behavior of -eucryptite can be attributed to three interdependent processes that all cause an increase in a but a decrease in c with increasing temperature: (i) Si/Al tetrahedral deformation, (ii) Li positional disordering, and (iii) tetrahedral tilting. Because disordered -eucryptite does not exhibit tetrahedral tilting, the absolute values of its axial thermal coefficients are smaller than those for the ordered sample. At low temperatures, both ordered and disordered -eucryptite exhibit a continuous expansion parallel to the c axis with decreasing temperature, whereas a remains approximately unchanged. Our difference Fourier synthesis reveals localization of Li ions below room temperature, and we suggest that repulsion between Li and Al/Si inhibits contraction along the a axes.
An organic-inorganic halide CH NH SnI perovskite with significantly improved structural stability is obtained via pressure-induced amorphization and recrystallization. In situ high-pressure resistance measurements reveal an increased electrical conductivity by 300% in the pressure-treated perovskite. Photocurrent measurements also reveal a substantial enhancement in visible-light responsiveness. The mechanism underlying the enhanced properties is shown to be associated with the pressure-induced structural modification.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.