Online communities such as Facebook and Twitter are enormously popular and have become an essential part of the daily life of many of their users. Through these platforms, users can discover and create information that others will then consume. In that context, recommending relevant information to users becomes critical for viability. However, recommendation in online communities is a challenging problem: 1) users' interests are dynamic, and 2) users are influenced by their friends. Moreover, the influencers may be context-dependent. That is, different friends may be relied upon for different topics. Modeling both signals is therefore essential for recommendations.We propose a recommender system for online communities based on a dynamic-graph-attention neural network. We model dynamic user behaviors with a recurrent neural network, and contextdependent social influence with a graph-attention neural network, which dynamically infers the influencers based on users' current interests. The whole model can be efficiently fit on large-scale data. Experimental results on several real-world data sets demonstrate the effectiveness of our proposed approach over several competitive baselines including state-of-the-art models. The source code and data are available at https://github.com/DeepGraphLearning/ RecommenderSystems.

No abstract

No abstract

In a neutral π-radical-based organic light-emitting diode (OLED), although the emission comes from the doublet excitons and their transition to the ground state is spin-allowed, the upper limit of internal quantum efficiency (IQE) is not clear, 50% or 100%? In this work, the deep-red OLEDs based on a neutral π-radical were fabricated. Up to 100% doublet exciton formation ratio was obtained through rational designing device structure and host-guest doping system. This indicates the IQE of neutral π-radical-based OLEDs will reach 100% if the nonradiative pathways of radicals can be suppressed. The maximum external quantum efficiency of the optimized device is as high as 4.3%, which is among the highest values of deep-red/near-infrared OLEDs with nonphosphorescent materials as emitters. Our results also indicate that using partially reduced radical mixture as emitter may be a way to solve aggregation-caused quenching in radical-based OLEDs.

The effects of the rigidity of molecular recognition sites in fluorene-based conjugated polymers P1 and P2 on metal ion sensing have been investigated. The structures of polymers P1 and P2 have twisted 2,2′-bipyridine and planar 1,10-phenanthroline units, respectively, which alternate with one fluorene monomer unit. It is found that the absorption and emission bands of 1,10-phenanthroline-based polymer P2 exposed to metal ions can be red-shifted up to 30 nm, and emission intensity can be quenched up to 100%, depending on metal ions present, which is very similar to that of the 2,2′-bipyridine-based analogue P1. However, polymer P2 shows much higher sensitivity to metal ions than P1. The origins of ionochromic effects of the 2,2′-bipyridinebased conjugated polymer due to the metal ion chelation have been attributed to both conformational changes and electron density variations on the polymer chains caused by introducing positively charged metal ions (Chen et al. J. Phys. Chem., B 2000, 104, 1950-1960. On the basis of the comparison of P2 with P1, conformational changes are not required in the ion responsive process of the phen ion-recognition unit. We demonstrate that the electron density variations play more important roles in metal ion-induced red-shifts in absorption and fluorescence quenching in photoluminescence.