Interpretable Multi-Head Self-Attention Architecture for Sarcasm Detection in Social Media

With the online presence of more than half the world population, social media plays a very important role in the lives of individuals as well as businesses alike. Social media enables businesses to advertise their products, build brand value, and reach out to their customers. To leverage these social media platforms, it is important for businesses to process customer feedback in the form of posts and tweets. Sentiment analysis is the process of identifying the emotion, either positive, negative or neutral, associated with these social media texts. The presence of sarcasm in texts is the main hindrance in the performance of sentiment analysis. Sarcasm is a linguistic expression often used to communicate the opposite of what is said, usually something that is very unpleasant, with an intention to insult or ridicule. Inherent ambiguity in sarcastic expressions make sarcasm detection very difficult. In this work, we focus on detecting sarcasm in textual conversations from various social networking platforms and online media. To this end, we develop an interpretable deep learning model using multi-head self-attention and gated recurrent units. The multi-head self-attention module aids in identifying crucial sarcastic cue-words from the input, and the recurrent units learn long-range dependencies between these cue-words to better classify the input text. We show the effectiveness of our approach by achieving state-of-the-art results on multiple datasets from social networking platforms and online media. Models trained using our proposed approach are easily interpretable and enable identifying sarcastic cues in the input text which contribute to the final classification score. We visualize the learned attention weights on a few sample input texts to showcase the effectiveness and interpretability of our model.     

Aqueous Precursor Induced Morphological Change and Improved Water Stability of CsPbBr3 Nanocrystals

In the literature, lead halide perovskites are very notable for their degradation in the presence of polar solvents, such as water. In contrast, in this research, it is observed that adding a minor amount of water into the precursor solution can improve the stability and photoluminescence quantum yield of CsPbBr₃ nanocrystals through a ligand-assisted reprecipitation (LARP) method. In this way, the shape and phase transformation from CsPbBr₃ nanoplates to CsPbBr₃/Cs₄PbBr₆ nanorods and Cs₄PbBr₆ nanowires can be controlled with increasing water content in the precursor solution. Upon adding water up to an ideal amount, CsPbBr₃ maintains its phase and nanoplate morphology. The key role of water amount for tuning the crystallinity, stability, morphology, optical properties, and phase transformation of cesium lead halide perovskite nanocrystals will be beneficial in the future commercialization of optoelectronics.     

Photochemical study of electrocyclization of 4-aryl-5-hetarylimidazolones for information optical recording

1. Download : Download high-res image (132KB)
2. Download : Download full-size image

     

Synthesis of Aluminophosphate Molecular Sieves in Alkaline Media

Unlike conventional aluminosilicate zeolites synthesized in alkaline media, aluminophosphate molecular sieves (AlPOs) have always been prepared under acidic conditions in the past three decades; this has been regarded as one of essential factors for synthesis, except for the case of silica-substituted analogues (SAPOs). For the first time, we demonstrate herein a simple and generalized route for synthesizing various types of aluminophosphate molecular sieves in alkaline media. A series of aluminophosphate sieves and their analogues have been prepared with different quaternary ammonium cations as structure-directing agents in this manner. The above successes have extended the systematic media from acidic or neutral to alkaline for the preparation of a series of aluminophosphate molecular sieves, which possibly open an alternative route for the synthesis of aluminophosphate molecular sieves.     

Toward the design of efficient adsorbents for Hg2+ removal: Molecular and thermodynamic insights

A systematic DFT study was performed to evaluate the effect of oxygenated functional groups for Hg²⁺ adsorption in aqueous systems. This work includes several aspects usually neglected in many current works, namely, ground-state multiplicity, solvation effects, establishment of thermodynamic parameters, atomic charge transfer, and modeling of infrared spectra. In addition, two carbonaceous models were studied to account for both the effect of the carbonaceous matrix and the oxygenated functional groups on the Hg²⁺ binding. Adsorption energies indicated that Hg²⁺ adsorption on the unsaturated model is favored in the following order: phenol > lactone > semiquinone > carboxyl, whereas for the saturated model, the Hg²⁺ adsorption energy decrease order is: carboxyl > semiquinone > lactone. Thermodynamic parameters confirmed that the adsorption process is spontaneous (unsaturated model), while the infrared spectra provided an insight at the atomic level about the experimentally reported bands. Our results contributed to a deeper understanding of the current experimental information on the effect of the surface functional groups on the Hg²⁺ adsorption over carbonaceous materials as different active sites can be present on oxygenated carbonaceous materials for metal adsorption. The results also create new ways to improve the performance of adsorption capability of mercury and other pollutants.     

Setting the Foundations of Aqueous Three-Phase Systems (A3PS) in the Quest for a Rational Design

This work presents the first in-depth study of Aqueous Three-Phase Systems (A3PS) with the main purpose of unveiling their behaviour, hence contributing to the development of this new field. Thus, a complete definition of a quaternary system was carried through by describing all the regions in detail to represent them later on in a regular-tetrahedral diagram. The three aqueous faces of the tetrahedron demonstrated an undeviating influence in the segregation capacity. Furthermore, a method for comparing Aqueous Biphasis Systems (ABS) immiscibilities was set up in order to allow the evaluation and detection of the “limiting ABS” for the three-phase region. Finally, all this information was compiled and utilised to obtain a new strategy for an A3PS rational design, which can be applied with ABS libraries or in an experimental approach. In this sense, this strategy represents an undoubted advance towards future studies and development of A3PS, as this sequential application of the constructed knowledge is assumed to save time and resources.     

Multi-Hollow Surface Dielectric Barrier Discharge for Bacterial Biofilm Decontamination

The plasma-activated gas is capable of decontaminating surfaces of different materials in remote distances. The effect of plasma-activated water vapor on Staphylococcus epidermidis, methicillin-resistant Staphylococcus aureus, Pseudomonas aeruginosa, and Escherichia coli biofilm contamination was investigated on the polypropylene nonwoven textile surface. The robust and technically simple multi-hollow surface dielectric barrier discharge was used as a low-temperature atmospheric plasma source to activate the water-based medium. The germicidal efficiency of short and long-time exposure to plasma-activated water vapor was evaluated by standard microbiological cultivation and fluorescence analysis using a fluorescence multiwell plate reader. The test was repeated in different distances of the contaminated polypropylene nonwoven sample from the surface of the plasma source. The detection of reactive species in plasma-activated gas flow and condensed activated vapor, and thermal and electrical properties of the used plasma source, were measured. The bacterial biofilm decontamination efficiency increased with the exposure time and the plasma source power input. The log reduction of viable biofilm units decreased with the increasing distance from the dielectric surface.