Search

Estimating Click-Through Rate (CTR) is a vital yet challenging task in personalized product search. However, existing CTR methods still struggle in the product search settings due to the following three challenges including how to more effectively extract users' short-term interests with respect to multiple aspects, how to extract and fuse users' long-term interest with short-term interests, how to address the entangling characteristic of long and short-term interests. To resolve these challenges, in this paper, we propose a new approach named Hierarchical Interests Fusing Network (HIFN), which consists of four basic modules namely Short-term Interests Extractor (SIE), Long-term Interests Extractor (LIE), Interests Fusion Module (IFM) and Interests Disentanglement Module (IDM). Specifically, SIE is proposed to extract user's short-term interests by integrating three fundamental interests encoders within it namely query-dependent, target-dependent and causal-dependent interest encoder, respectively, followed by delivering the resultant representation to the module LIE, where it can effectively capture user long-term interests by devising an attention mechanism with respect to the short-term interests from SIE module. In IFM, the achieved long and short-term interests are further fused in an adaptive manner, followed by concatenating it with original raw context features for the final prediction result. Last but not least, considering the entangling characteristic of long and short-term interests, IDM further devises a self-supervised framework to disentangle long and short-term interests. Extensive offline and online evaluations on a real-world e-commerce platform demonstrate the superiority of HIFN over state-of-the-art methods., Comment: accpeted by CIKM'22 as a Full Paper

The phase behavior of hydrocarbon fluids confined in porous media has been reported to deviate significantly from that in the bulk environment due to the existence of sub-10nm pores. Though experiments and simulations have measured the bubble/dew points and sorption isotherms of hydrocarbons confined in both natural and synthetic nanopores, the confinement effects in terms of the strength of fluid-pore interactions tuned by surface wettability and chemistry have received comparably less discussion. More importantly, the underlying physics of confinement-induced phenomena remain obfuscated. In this work, we studied the phase behavior and capillary condensation of n-hexane to understand the effects of confinement at the molecular level. To systematically investigate the pore effects, we constructed two types of wall confinements; one is a structureless virtual wall described by the Steele potential and the other one is an all-atom amorphous silica structure with surface modified by hydroxyl groups. Our numerical results demonstrated the importance of fluid-pore interaction, pore size, and pore morphology effects in mediating the pressure-volume-temperature (PVT) properties of hydrocarbons. The most remarkable finding of this work was that the saturation pressure predicted from the van der Waals-type adsorption isothermal loop could be elevated or suppressed relative to the bulk phase, as illustrated in the graphical abstract. As the surface energy (i.e., fluid-pore interaction) decreased, the isothermal vapor pressure increased, indicating a greater preference for the fluid to exist in the vapor state. Sufficient reduction of the fluid-pore interactions could even elevate the vapor pressure above that of the bulk fluid.

Accurate and efficient flow models for hydrocarbons are important in the development of enhanced geotechnical engineering for energy source recovery and carbon capture & storage in low-porosity, low-permeability rock formations. This work reports an atomistically-validated, mesoscopic model for heptane based on a many-body dissipative particle dynamics (mDPD) method. In this model, each heptane molecule is coarse-grained in one mDPD bead and the mDPD model parameters are calibrated with a rigorous approach using reference data, including experimental measurements and/or molecular dynamics (MD) simulations. Results show that this mDPD model accurately predicts the bulk pressure-density relation of heptane and surface tension. Notice that our approach can be used to calibrate the mDPD model for other hydrocarbons as well, though heptane is chosen as a representative source fluid for its abundance in source rocks. Further, our timing test indicates that the mDPD model is three orders of magnitude faster than its MD counterpart for simulations of bulk heptane in equivalent volumes. Overall, this work serves as a key prerequisite for the development of accurate and efficient mesoscale models for the flow of hydrocarbons confined in mesoporous rock formations., Comment: 16 pages, 7 figures