Search

As a country with the largest population in the world, China has a large number of school-age children every year. The enrolment of these children deserves attention. Taking two typical areas of Beijing and Shenzhen as examples, this paper explores the internal causes and solutions of educational inequality by analyzing typical policies such as "high-priced school district housing" and "point admissions"

Effective detection of antibiotics and photodegradation using a novel zinc-based metal–organic framework.

Session-based recommendation (SBR) , which mainly relies on a user’s limited interactions with items to generate recommendations, is a widely investigated task. Existing methods often apply RNNs or GNNs to model user’s sequential behavior or transition relationship between items to capture her current preference. For training such models, the supervision signals are merely generated from the sequential interactions inside a session, neglecting the correlations of different sessions, which we argue can provide additional supervisions for learning the item representations. Moreover, previous methods mainly adopt the cross-entropy loss for training, where the user’s ground truth preference distribution towards items is regarded as a one-hot vector of the target item, easily making the network over-confident and leading to a serious overfitting problem. Thus, in this article, we propose a Collaborative Graph Learning (CGL) approach for session-based recommendation. CGL first applies the Gated Graph Neural Networks (GGNNs) to learn item embeddings and then is trained by considering both the main supervision as well as the self-supervision signals simultaneously. The main supervisions are produced by the sequential order while the self-supervisions are derived from the global graph constructed by all sessions. In addition, to prevent overfitting, we propose a Target-aware Label Confusion (TLC) learning method in the main supervised component. Extensive experiments are conducted on three publicly available datasets, i.e., Retailrocket, Diginetica, and Gowalla. The experimental results show that CGL can outperform the state-of-the-art baselines in terms of Recall and MRR.

COVID-19 and the pandemic-induced lockdowns juxtaposed against the surge in the number of college graduates have made the dilemma of “fierce competition and difficult employment” more real. The employment of college students has become a topic of serious concern in society. This study aimed to develop a Future Employability Scale for Chinese college students and evaluate its reliability and validity. Based on the analysis of the literature, the study developed the initial measurement scale of the college students’ future employability and calibrated the initial measurement and question volume based on experts’ feedback. First, the students’ group was measured, and data from 389 university students were collected and analyzed. Second, the data collection and verification factor analysis of 387 university students were collected and verified, and the internal consistency reliability, split-half reliability, and validity of the scale were evaluated. Further, 68 college students were selected to evaluate their test-retest reliability after an interval of one month. The Future Employability Scale of college students had 28 items covering four dimensions: knowledge skill, personality quality, interpersonal network, and career development. The reliability test found that the total scale of the Future Employability Scale and the internal consistency reliability, split-half reliability, and retest reliability of each dimension were good, and the validity test suggested that the scale had good content validity, structural validity, and calibration correlation validity. With a clear structure, good reliability, and validity, the Future Employability Scale is a good tool to measure the future employability of college students.

The existence of antibiotics in the environment can trigger a number of issues by fostering the widespread development of antimicrobial resistance. Currently, the most popular techniques for removing antibiotic pollutants from water include physical adsorption, flocculation, and chemical oxidation, however, these processes usually leave a significant quantity of chemical reagents and polymer electrolytes in the water, which can lead to difficulty post-treating unmanageable deposits. Furthermore, though cost-effectiveness, efficiency, reaction conditions, and nontoxicity during the degradation of antibiotics are hurdles to overcome, a variety of photocatalysts can be used to degrade pollutant residuals, allowing for a number of potential solutions to these issues. Thus, the urgent need for effective and rapid processes for photocatalytic degradation leads to an increased interest in finding more sustainable catalysts for antibiotic degradation. In this review, we provide an overview of the removal of pharmaceutical antibiotics through photocatalysis, and detail recent progress using different nanostructure-based photocatalysts. We also review the possible sources of antibiotic pollutants released through the ecological chain and the consequences and damages caused by antibiotics in wastewater on the environment and human health. The fundamental dynamic processes of nanomaterials and the degradation mechanisms of antibiotics are then discussed, and recent studies regarding different photocatalytic materials for the degradation of some typical and commonly used antibiotics are comprehensively summarized. Finally, major challenges and future opportunities for the photocatalytic degradation of commonly used antibiotics are highlighted.

Background and Purpose: Irritable bowel syndrome (IBS) is usually associated with chronic gastrointestinal disorders. Its most common subtype is accompanied with diarrhea (IBS-D). The enteric nervous system (ENS) modulates major gastrointestinal motility and functions whose aberration may induce IBS-D. The enteric neurons are susceptible to long-term neurotransmitter level alterations. The patchouli alcohol (PA), extracted from Pogostemonis Herba, has been reported to regulate neurotransmitter release in the ENS, while its effectiveness against IBS-D and the underlying mechanism remain unknown.Experimental Approach: In this study, we established an IBS-D model in rats through chronic restraint stress. We administered the rats with 5, 10, and 20 mg/kg of PA for intestinal and visceral examinations. The longitudinal muscle myenteric plexus (LMMP) neurons were further immunohistochemically stained for quantitative, morphological, and neurotransmitters analyses.Key Results: We found that PA decreased visceral sensitivity, diarrhea symptoms and intestinal transit in the IBS-D rats. Meanwhile, 10 and 20 mg/kg of PA significantly reduced the proportion of excitatory LMMP neurons in the distal colon, decreased the number of acetylcholine (Ach)- and substance P (SP)-positive neurons in the distal colon and restored the levels of Ach and SP in the IBS-D rats.Conclusion and Implications: These findings indicated that PA modulated LMMP excitatory neuron activities, improved intestinal motility and alleviated IBS-induced diarrheal symptoms, suggesting the potential therapeutic efficacy of PA against IBS-D.

In order to develop a simple and efficient site-directed mutagenesis solution, the Gibson assembly technique was used to clone the cyclin dependent kinase 4 gene with single or double site mutations, with the aim to simplify the overlap extension PCR. The gene fragments containing site mutations were amplified using a strategy similar to overlap extension PCR. Meanwhile, an empty plasmid was digested by double restriction endonucleases to generate a linearized vector with a short adaptor overlapping with the targeted gene fragments. The gene fragments were directly spliced with the linearized vector by Gibson assembly in an isothermal, single-reaction, creating a recombinant plasmid. After the recombinant plasmids were transformed into competent

Most ECDs are coated with an electrochromic material on the transparent conductive oxide (TCO) substrate. A novel electrochromic device (ECD), having a variable optical performance, was prepared by using tungsten foil as a substrate in this study. It was found that the WO

Cementitious materials are the most widely used in the construction. However, cementitious materials have defects such as low tensile strength which is easy to cause the formation of cracks. In recent years, with the research and application of smart materials, intelligent cementitious materials with self-sensing, self-diagnosis and self-healing functions have received extensive researcher’s attention. In this paper, a polyacrylic hydrogel that could be utilized for self-healing of cementitious materials was designed, and the three-dimensional network structure of the hydrogel controlled the release rate of encapsulated repair agent. The phosphate-incorporated hydrogel was filled into cement slurry to prepare self-healing cementitious materials. After the cement sample cracked and water penetrated into the crack, the phosphate was released from hydrogel. Phosphate reacted with the calcium in the pore solution to form hydroxyapatite type minerals which healed the crack. The self-healing cementitious material is in a position to heal cracks of 300 μm in width. After healing for 28 days at 20oC and 95% of RH, the compressive strength of precracked specimens could reach 85% of that of intact ones.

Basket-Sensitive Item Recommendation (BSIR) is a challenging task that aims to recommend an item to add to the current basket given a user’s historical behaviors. The recommended item is supposed to be relevant to the items in current basket. Previous works mainly produce a recommendation based on user’s current basket, ignoring the inherent preference released by user’s long-term behaviors and failing to accurately distinguish the item importance in the basket for detecting user intent. To tackle the above issues, we propose a hybrid model, i.e., Hybrid-Preference Neural Model (HPNM), where a user’s inherent preference is recognized by modeling the historical sequential baskets and the recent preference is identified by focusing on the current basket. In detail, we apply an attention mechanism for distinguishing the importance of items in a basket to generate an accurate basket representation. GRU is utilized for modeling the basket-level sequential information to obtain user’s long-term preference and the representation of the current session is regarded as user’s short-term preference. We evaluate the performance of our proposals against the state-of-the-art baselines in the field of BSIR on two public datasets, i.e., TaFeng and Foursquare. The experimental results show that HPNM can achieve obvious improvements against the baselines in terms of HLU and Recall. In addition, we find HPNM with an attention mechanism can lead to a larger improvement against the baseline for item recommendation in terms of HLU and Recall on testing baskets with relatively fewer items.

Here, we present a high repetition rate, narrow bandwidth, extreme ultraviolet photon source for time- and angle-resolved photoemission spectroscopy. The narrow bandwidth pulses

Ultrafast light sources have become an indispensable tool to access and understand transient phenomenon in material science. However, a simple and easy-to-implement method for harmonic selection, with high transmission efficiency and pulse duration conservation, is still a challenge. Here we showcase and compare two approaches for selecting the desired harmonic from a high harmonic generation source while achieving the above goals. The first approach is the combination of extreme ultraviolet spherical mirrors with transmission filters and the second approach uses a normal-incidence spherical grating. Both solutions target time- and angle-resolved photoemission spectroscopy with photon energies in the 10-20 eV range but are relevant for other experimental techniques as well. The two approaches for harmonic selection are characterized in terms of focusing quality, photon flux, and temporal broadening. It is demonstrated that a focusing grating is able to provide much higher transmission as compared to the mirror+filter approach (3.3 times higher for 10.8 eV and 12.9 times higher for 18.1 eV), with only a slight temporal broadening (6.8% increase) and a somewhat larger spot size (∼30% increase). Overall, our study establishes an experimental perspective on the trade-off between a single grating normal incidence monochromator design and the use of filters. As such, it provides a basis for selecting the most appropriate approach in various fields where an easy-to-implement harmonic selection from high harmonic generation is needed.

Label representation aims to generate a so-called verbalizer to an input text, which has a broad application in the field of text classification, event detection, question answering, etc. Previous works on label representation, especially in a few-shot setting, mainly define the verbalizers manually, which is accurate but time-consuming. Other models fail to correctly produce antonymous verbalizers for two semantically opposite classes. Thus, in this paper, we propose a metric sentiment learning framework (MSeLF) to generate the verbalizers automatically, which can capture the sentiment differences between the verbalizers accurately. In detail, MSeLF consists of two major components, i.e., the contrastive mapping learning (CML) module and the equal-gradient verbalizer acquisition (EVA) module. CML learns a transformation matrix to project the initial word embeddings to the antonym-aware embeddings by enlarging the distance between the antonyms. After that, in the antonym-aware embedding space, EVA first takes a pair of antonymous words as verbalizers for two opposite classes and then applies a sentiment transition vector to generate verbalizers for intermediate classes. We use the generated verbalizers for the downstream text classification task in a few-shot setting on two publicly available fine-grained datasets. The results indicate that our proposal outperforms the state-of-the-art baselines in terms of accuracy. In addition, we find CML can be used as a flexible plug-in component in other verbalizer acquisition approaches.

Membrane technology has received wide attention as it plays an important role in water treatment process. However, membrane fouling results in a decrease in water permeation flux and separation effect, which lead to degraded membrane lifespan. A facile and novel strategy to form a multifunctional modified coating on the membrane surface by using the one-step assembly coating on the basis of tannin acid had been reported in this work. Tannin acid is a plant-derived polyphenolic substance that is hydrophilic and bactericidal. As a common nonionic polymer, polyvinylpyrrolidone (PVP) has a hydrophobic interaction and hydrogen bonding with TA and is beneficial to improve the hydrophilicity and antifouling performance of the substrate. However, the benefits of coating on the membrane have been limited due to the strong complexing ability of TA and PVP in aqueous solution. Herein, 50% N,N-dimethylformamide (DMF) aqueous solution were found that promoted the co-deposition of TA and PVP on the surface of the membrane. The catechol-rich polyphenol coating provides versatile properties on the membrane surface, including ultra-high pure water flux, high antifouling properties, recyclability, and a certain degree of antimicrobial performance. Moreover, the TA-PVP coating imparts a binding site for binding metal ions that allows it to capture metal ions through complexation and immobilization on the coating. The modified membranes showed more significant antifouling performance after iron ions have been complexed, and showed stronger bactericidal activity after silver ions have been complexed.

Text representation, which converts text spans into real-valued vectors or matrices, is a crucial tool for machines to understand the semantics of text. Although most previous works employed classic methods based on statistics and neural networks, such methods might suffer from data sparsity and insensitivity to the text structure, respectively. To address the above drawbacks, we propose a general and structure-sensitive framework, i.e., the hierarchical architecture. Specifically, we incorporate the hierarchical architecture into three existing neural network models for document representation, thereby producing three new representation models for document classification, i.e., TextHFT, TextHRNN, and TextHCNN. Our comprehensive experimental results on two public datasets demonstrate the effectiveness of the hierarchical architecture. With a comparable (or substantially less) time expense, our proposals obtain significant improvements ranging from 4.65 to 35.08% in terms of accuracy against the baseline. We can conclude that the hierarchical architecture can enhance the classification performance. In addition, we find that the benefits provided by the hierarchical architecture can be strengthened as the document length increases.

In this study, a novel ionically crosslinked gel polymer electrolyte (PADA gel electrolyte) with a transmittance of more than 97% was successfully fabricated and applied in electrochromic devices (ECDs) as an ion transport layer. The PADA gel-based ECDs (PECDs) exhibited the large optical modulation of 61% at 660 nm, the high coloration efficiency of 78.7 cm2 C−1, and a good memory effect. The shortest colored time (tc) and bleached time (tb) of the PECD were 7.5 s and 8.5 s, respectively, due to high Li+ mobility of the PADA layer. The ionic conductivities of the PADA gel electrolyte are 1.33 × 10−2 S cm−1 and 0.56 × 10−2 S cm−1 at 25 °C and −10 °C respectively, since its conductive behavior is similar to that of liquid electrolytes. The PECDs showed good cycling durability and long-term stability, which is attributed to the stable structures of the PADA gel electrolyte.

Event detection classifies unlabeled sentences into event labels, which can benefit numerous applications, including information retrieval, question answering and script learning. One of the major obstacles to event detection in reality is insufficient training data. To deal with the low-resources problem, we investigate few-shot event detection in this paper and propose TaLeM, a novel taxonomy-aware learning model, consisting of two components, i.e., the taxonomy-aware self-supervised learning framework (TaSeLF) and the taxonomy-aware prototypical networks (TaPN). Specifically, TaSeLF mines the taxonomy-aware distance relations to increases the training examples, which alleviates the generalization bottleneck brought by the insufficient data. TaPN introduces the Poincare embeddings to represent the label taxonomy, and integrates them into a task-adaptive projection networks, which tackles problems of the class centroids distribution and the taxonomy-aware embedding distribution in the vanilla prototypical networks. Extensive experiments in the four types of meta tasks demonstrate the superiority of our proposal over the strong baselines, and further verify the effectiveness and importance of modeling the label taxonomy. Besides, TaSeLF can be a flexible plug-in for the other taxonomy-based few-shot classification tasks.

Chloride ions in the seaside environment can corrode the steel reinforcement in concrete, which greatly endangers the safety of seaside structures. As an excellent adsorption material, hydrogel is widely used in the field of water treatment but is rarely used in cementitious materials. In this study, a polyacrylamide–chitosan hydrogel (PAMC) was prepared with N,N-methylenebisacrylamide as the cross-linking agent and acrylamide as the monomer. The prepared PAMC gel could effectively adsorb chloride ions in simulated seawater and simulated sea sand environments, and the maximum adsorption capacity of chloride ions by PAMC-1 (prepared from 2.5 g acrylamide and 1% content of N,N-methylenebisacrylamide relative to acrylamide) gels in simulated seawater was 55.53 mg/g. The adsorption behavior of the PAMC gels in solution fit the Langmuir isotherm model. The composition and morphology of the PAMC gel were characterized, and the responsiveness of the PAMC gel to the environment was studied. The results showed that the PAMC gels adsorbed better in alkaline environments and thus could be used in alkaline cement-based environments. The mortar sample containing the PAMC-1 gel had higher resistance to chloride ion penetration, and the chloride ion content at 7.5–10mm from the surface of the sample cured for 28 days was reduced by 41.4% compared to the samples without the gel.

Session-based recommendation is a challenging task. Without access to a user's historical user-item interactions, the information available in an ongoing session may be very limited. Previous work on session-based recommendation has considered sequences of items that users have interacted with sequentially. Such item sequences may not fully capture complex transition relationship between items that go beyond inspection order. Thus graph neural network (GNN) based models have been proposed to capture the transition relationship between items. However, GNNs typically propagate information from adjacent items only, thus neglecting information from items without direct connections. Importantly, GNN-based approaches often face serious overfitting problems. We propose Star Graph Neural Networks with Highway Networks (SGNN-HN) for session-based recommendation. The proposed SGNN-HN applies a star graph neural network (SGNN) to model the complex transition relationship between items in an ongoing session. To avoid overfitting, we employ highway networks (HN) to adaptively select embeddings from item representations. Finally, we aggregate the item embeddings generated by the SGNN in an ongoing session to represent a user's final preference for item prediction. Experiments on two public benchmark datasets show that SGNN-HN can outperform state-of-the-art models in terms of P@20 and MRR@20 for session-based recommendation.

A commercial PVDF Microfiltration (MF) membrane was surface modified via a simple coating method for improvement of the hydrophilicity and anti-fouling performance. Herein, trace amounts of tannin acid (TA) and Polyethyleneimine (PEI) were firstly used with (3-Chloropropyl)trimethoxysilan (CTS) to endow the PVDF membranes with hydrophilicity. The physicochemical property of the modified membranes was characterized by SEM, AFM, ATR-FTIR and XPS respectively, and a series of tests including water contact angle (WCA), underwater oil contact angle (OCA), pure water flux (PWF), anti-fouling experiments and so on were utilized to inspect the modified effect. Benefiting from the interactions among CTS, PEI and TA, several coating layers formed on the surface of the membranes and remarkable hydrophilicity with water contact angle of 16° was obtained, moreover, the pure water flux of this composite membranes could reach 10,782 L/m2·h.

Query suggestions help users refine their queries after they input an initial query. Previous work on query suggestion has mainly concentrated on approaches that are similarity-based or context-based, developing models that either focus on adapting to a specific user (personalization) or on diversifying query aspects in order to maximize the probability of the user being satisfied (diversification). We consider the task of generating query suggestions that are both personalized and diversified. We propose a personalized query suggestion diversification (PQSD) model, where a user’s long-term search behavior is injected into a basic greedy query suggestion diversification model that considers a user’s search context in their current session. Query aspects are identified through clicked documents based on the open directory project (ODP) with a latent dirichlet allocation (LDA) topic model. We quantify the improvement of our proposed PQSD model against a state-of-the-art baseline using the public america online (AOL) query log and show that it beats the baseline in terms of metrics used in query suggestion ranking and diversification. The experimental results show that PQSD achieves its best performance when only queries with clicked documents are taken as search context rather than all queries, especially when more query suggestions are returned in the list.

A major challenge in applying Model Predictive Control (MPC) to building automation and control (BAC) is the development of a simplified mathematical model of the building for real-time control with fast response times. However, building models are highly complex due to nonlinearities in heat and mass transfer processes of the building itself and the accompanying air-conditioning and mechanical ventilation systems. This paper proposes a method to develop an integrated state-space model (SSM) for indoor air temperature, radiant temperature, humidity and Predicted Mean Vote (PMV) index suitable for fast real-time multiple objectives optimization. Using the model, a multi-objective MPC controller is developed and its performance is evaluated through a case study on the BCA SkyLab test bed facility in Singapore. The runtime of the MPC controller is less than 0.1 s per optimization, which is suitable for real-time BAC applications. Compared to the conventional ON/OFF control, the MPC controller can achieve up to 19.4% energy savings while keeping the PMV index within the acceptable comfort range. When the MPC controller is adjusted to be thermal-comfort-dominant that achieves a neutral PMV index at most office hours, the system can still bring about 6% in energy savings as compared to the conventional ON/OFF control. NRF (Natl Research Foundation, S’pore) Accepted version

The storage of living cells is the major challenge for cell research and cell treatment. Here, we introduced a novel supramolecular gel cryopreservation system which was prepared in the microchannel, and the supramolecular gel (BDTC) was self-assembled by gelator Boc- O-dodecyl-l-tyrosine (BDT). This cryopreservation system could obviously minimize the cell injury because the BDTC supramolecular gel had a more compact three-dimensional network structure when the BDT gelator self-assembled in the confined space of microchannel. This compact structure could confine the growth of the ice crystal, reduce the change rate of cell volumes and osmotic shock, decrease the freezing point of the cryopreservation system, and possess better protection capability. Furthermore, the results of functionality assessments showed that the thawed cells could grow and proliferate well and remain the same growth trend of the fresh cells after the RSC96 cells flowed out from the microchannel. This novel method has potential to be used for the cryopreservation of cells, cell therapy, and tissue engineering.

Totem, as the national symbol, reflects a nation’s culture, spirit and traditional values. This paper, basing on a lot of reading on Sino-US cultural differences, will explore the Sino-US cultural differences from a totally new way, the different characteristics of the totems of the two countries—the Chinese dragon and the American Eagle, and the different values, ways of thinking and cultural traditions shown by their totems respectively. This paper not only reveals the cultural differences between the two countries, more importantly, from the outside phenomenon to the inside nature, it finds out the underlying causes of these differences through comparative analysis. Hoping to enhance the mutual understanding between the two peoples and show the world that the Chinese people has always been a peace-loving nation since ancient time under the current situation that the Western countries strongly play up the theory of “China threat”.

Fact verification aims to retrieve related evidence from raw text to verify the correctness of a given claim. Existing works mainly leverage the single-granularity features for the representation learning of evidences, i.e., sentence features, ignoring other features like entity-level and context-level features. In addition, they usually focus on improving the prediction accuracy while lacking the interpretability of the inference process, which leads to unreliable results. Thus, in this paper, to investigate how to utilize multi-granularity semantic units for evidence representation as well as to improve the explainability of evidence reasoning, we propose a Hierarchical Reasoning-based Heterogeneous Graph Neural Network for fact verification (HHGN). HHGN combines multiple features of entity, sentence as well as context for evidence representation, and employs a heterogeneous graph to capture their semantic relations. Inspired by the human inference process, we design a hierarchical reasoning-based node updating strategy to propagate the evidence features. Then, we extract the potential reasoning paths from the graph to predict the label, which aggregates the results of different paths weighted by their relevance to the claim. We evaluate our proposal on FEVER, a large-scale benchmark dataset for fact verification. Our experimental results demonstrate the superiority of HHGN over the competitive baselines in both single evidence and multiple evidences settings. In addition, HHGN presents reasonable interpretability in the form of aggregating the features of relevant entity units and selecting the evidence sentences with high confidence.

Several studies indicate that human exposure to plasticizers via dermal pathway is not negligible, but the dermal bioaccessibility of phthalates and alternative plasticizers from the important environmental matrix including indoor dust and clothing and the importance weight of dermal exposure to those pollutants have been poorly studied. An in vitro physiologically based extraction test was employed to investigate the dermal bioaccessibility of target phthalates and alternative plasticizers from indoor dust and clothing. Temperature, incubation time, sweat/sebum ratio and solid/liquid ratio were selected to study their effects on the bioaccessibility. The bioaccessibility of Diethyl phthalates (DEP), dibutyl phthalate (DBP), bis-2-ethylhexyl phthalate (DEHP), Acetyl tributyl citrate (ATBC), bis-2-ethylhexyladipate (DEHA) and bis-2-ethylhexyl terephthalate (DEHT) in indoor dust were 66.20 ± 1.93%, 94.27 ± 1.31%, 80.37 ± 8.09%, 75.02 ± 2.12%, 94.50 ± 3.42% and 74.09 ± 3.79%, respectively, under the condition of 1:1 sweat/sebum ratio, 1/100 solid/liquid ratio (indoor dust), 1:1 area/area ratio (1:1, clothing) and 90 min incubation time at 36.3 °C which are chosen based on the experimental results and human physical conditions. DBP showed the highest bioaccessibility in all samples. The time course of the plasticizer release was fitted to a first-order one-compartment model. DBP showed the highest release rate (k1) calculated from the model, which was consistent with the bioaccessibility result. Risk assessment indicated that dermal exposure of DBP was an important exposure route, accounting for about 21.58% of total intake, and indoor dust was an important exposure media when considering the dermal bioaccessibility.

The adoption of model predictive control (MPC) for building automation and control applications is challenged by the high hardware and software requirements to solve its optimization problem. This study proposes an approximate MPC that mimics the dynamic behaviours of MPC using the recurrent neural network with a structure of nonlinear autoregressive network with exogenous inputs. The approximate MPC is developed by learning from the measured operation data of buildings controlled by MPC, therefore it can produce MPC-like control for buildings without needing to solve the optimization problem, significantly reducing the computation load as compared to MPC. The proposed approximate MPC is implemented in two testbeds, an office and a lecture theatre, to control the air-conditioning systems. The control performance of the approximate MPC is compared to MPC as well as the original reactive control of the two testbeds. The approximate MPC retained most of the energy and thermal comfort performance of MPC in both testbeds. For the office, the MPC and approximate MPC reduced 58.5% and 51.6% of cooling energy consumption, respectively, as compared to the original control. For the lecture theatre, the MPC and approximate MPC reduced 36.7% and 36.2% of cooling energy consumption, respectively, as compared to the original control. Meanwhile, both approximate MPC and MPC significantly improved indoor thermal comfort in the two testbeds as compared to their original control. Despite having minor degradation in control performance the approximate MPC was more than 100 times faster than MPC in generating optimal control commands in each time step. National Research Foundation (NRF) This research is financially supported by JTC Corporation (contract nos. N190107T00 and 2019-0607) and Smart Nation & Digital Government Office (SNDGO) of Singapore (Grant no. NRF2016IDM-TRANS001-031).

Concrete is liable to form cracks owing to the limited tensile strength, which prompts researchers to propose a concept of self-healing. Inspired by the biological bones self-healing, this paper utilized hydroxyapatite as healing substances of cement-based materials and designed polyacrylamide (PAM) gel microparticles as carriers of phosphate for cement paste self-healing system. Morphology and properties of PAM gel microparticles, feasibility of hydroxyapatite as healing agents and healing efficiency were studied. The results showed that the three-dimensional crosslinked structure of PAM can encapsulate phosphate and the content of crosslinking agent can control the release rate and cumulative release of phosphate. Chloride permeability of cement paste containing PAM with phosphate is reduced after self-healing compared with plain cement paste. The healing efficiency of self-healing cement-based materials can reach 81.9% for compressive strength repair ratio and 93.2% for resistance to water penetration repair ratio after curing for 28 days. The white reaction products at the repairing section are proved to be hydroxyapatite according to multiple analytical techniques.

Incorporating topic information can help response generation models to produce informative responses for chat-bots. Previous work only considers the individual semantic of each topic, ignoring its specific dialog context, which may result in inaccurate topic representation and hurt response coherence. Besides, as an important feature of multi-turn conversation, dynamic topic transitions have not been well-studied. We propose a Context-Controlled Topic-Aware neural response generation model, i.e., CCTA, which makes dialog context interact with the process of topic representing and transiting to achieve balanced improvements on response informativeness and contextual coherence. CCTA focuses on capturing the semantical relations within topics as well as their corresponding contextual information in conversation, to produce context-dependent topic representations at the word-level and turn-level. Besides, CCTA introduces a context-controlled topic transition strategy, utilizing contextual topics to yield relevant transition words. Extensive experimental results on two benchmark multi-turn conversation datasets validate the superiority of our proposal on generating coherent and informative responses against the state-of-the-art baselines. We also find that topic transition modeling can work as an auxiliary learning task to boost the response generation.

A model predictive control system with adaptive machine-learning-based building models for building automation and control applications is proposed. The system features an adaptive machine-learning-based building dynamics modelling scheme that updates the building model regularly using online building operation data through a dynamic artificial neural network with a nonlinear autoregressive exogenous structure. The system also employs a multi-objective function that could optimize both energy efficiency and indoor thermal comfort, two often contradicting demands. The proposed model predictive control system is implemented to control the air-conditioning and mechanical ventilation systems in two single-zone testbeds, an office and a lecture theatre, located in Singapore for experimental evaluation of its control performance. The model predictive control system is compared against the original reactive control system (thermostat in the office and building management system in the lecture theatre) in each testbed. The model predictive control system reduces 58.5% cooling thermal energy consumption in the office and 36.7% cooling electricity consumption in the lecture theatre, as compared to their respective original control. Meanwhile, the indoor thermal comfort in both testbeds is also greatly improved by the model predictive control system. Developing a model predictive control system using machine-learning-based building dynamics models could largely cut down the model construction time to days as compared to its counterpart using physics-based models, which usually take months to construct. However, the machine-learning-based modelling approach could be challenged by lack of building operational data necessary for model training in case of model predictive control development before the building has become operational. Nanyang Technological University This research is financially supported by Energy Research Institute at NTU (ERI@N), JTC Corporation (contract nos. N190107T00 and 2019-0607) and Smart Nation & Digital Government Office (SNDGO) of Singapore (Grant nos. NRF2016IDM-TRANS001-031)

Query auto-completion (QAC) is one of the most visible features in modern search engines. It helps users complete their queries by presenting a list of possible completions while they are typing in the search box. Existing works on QAC focus on employing learning-to-rank algorithms over handcrafted features. However, those manually designed features are unable to capture non-linear relationships between users and their submitted queries. Meanwhile, although Recurrent Neural Networks (RNNs) show significant advances in various areas, little attention is paid to its application to QAC. To bridge this gap, we propose three RNN-based models for QAC ranking: a simple session-based RNN model, a personalized RNN model and an attentive RNN model. Extensive experiments are conducted on a real-world query log. The significant improvement over the compared baseline verifies the effectiveness of the personalized RNN model and the attentive RNN model.

Query suggestions help users of a search engine to refine their queries. Previous work on query suggestion has mainly focused on incorporating directly observable features such as query co-occurrence and semantic similarity. The structure of such features is often set manually, as a result of which hidden dependencies between queries and users may be ignored. We propose an AHNQS model that combines a hierarchical structure with a session-level neural network and a user-level neural network to model the short- and long-term search history of a user. An attention mechanism is used to capture user preferences. We quantify the improvements of AHNQS over state-of-the-art RNN-based query suggestion baselines on the AOL query log dataset, with improvements of up to 21.86% and 22.99% in terms of MRR@10 and Recall@10, respectively, over the state-of-the-art; improvements are especially large for short sessions.

The blooming change of requirements poses a challenge for change impact analysis (CIA), especially in large scale software development. Existing CIA techniques exploiting manual analysis and automatic methods suffer from high cost, low accuracy and expert dependence in current industrial practice. Therefore, this paper proposes a fully automatic change impact analysis (FACIA) method to overcome the aforementioned shortcomings. We consider the requirement changes generally happen as the form of phrases. The impact units for the changed phrases may contain the changed phrases (CP1), the co-occurrence phrases of CP1(CP2), the tokens of CP1 in the changed requirement, the similar phrases of CP1 and the similar phrases of CP2 in the whole requirements. Then five algorithms are designed based on the combinations of those impact units, so as to get a more precise change propagation and generate the lists of change impact automatically. We conduct extensive evaluations for the proposed approach with two different industrial data sets. The results show that compared with the method of expert depended, our approach can get a reliable sorted list for CIA more quickly with lower cost.

Query suggestions help users refine their queries after they input an initial query. We consider the task of generating query suggestions that are personalized and diversified. We propose a personalized query suggestion diversification model (PQSD), where a user's long-term search behavior is injected into a basic greedy query suggestion diversification model (G-QSD) that considers a user's search context in their current session. Query aspects are identified through clicked documents based on the Open Directory Project (ODP). We quantify the improvement of PQSD over a state-of-the-art baseline using the AOL query log and show that it beats the baseline in terms of metrics used in query suggestion ranking and diversification. The experimental results show that PQSD achieves the best performance when only queries with clicked documents are taken as search context rather than all queries.

In order to improve the thermal energy utilization efficiency of aircraft, an integrated thermal management system model using endothermic fuel has been created in AMESIM. The system includes several thermal subsystems such as environment control system, the hydraulic equipment and power equipment cooling system, and anti-icing/de-icing system. The fuel oil was chosen for heat sink and it could cool the exothermic thermal system and transmit the energy to anti-icing/de-icing system in case of icing. Several simulation examples including three atmospheric conditions were accomplished. The results indicated that the present system had the ability to reuse the waste heat meanwhile all the thermal system worked normally. The anti-icing/de-icing didn’t need additional energy which realized less energy consumption. In addition, the model has the ability to facilitate the analysis and optimization of thermal systems on board.

A novel bone-mimic self-healing cementitious material was developed by mimicking the healing process of bionic bone fracture using hydrogel impregnated with phosphate in Portland cement paste. The properties of phosphate-incorporated hydrogel, feasibility of hydroxyapatite formation in pore solution, release of phosphate from the hydrogel into cement paste, phase assemblage of self-healing products in the crack of cement paste and the self-healing efficiency were investigated with a range of analytical techniques. The phosphate-incorporated hydrogel can release phosphate into cracks at controlled rate. Carbonated and calcium deficient hydroxyapatite particles with sizes of approximately 30 μm were found as the main phase assemblage in the cracks during the self-healing process. The healing products grew from the surface of both sides to the center of crack. The compressive strength and impermeability of the self-healing cement pastes containing hydrogel impregnated with phosphates were fully restored after being cured for 28 days. The autonomous self-healing by introducing phosphate in hydrogel contributed the most to healing capacity, followed by the autogenous self-healing driven by the water released from hydrogel and the autogenous self-healing of cement paste. The hydroxyapatite-type products intermixed with minor amounts of calcite formed in the cracks accompanying the controlled release of phosphate from the hydrogel, providing the self-healing capabilities of the cement paste.

Model predictive control (MPC) in building automation and control (BAC) applications is challenged by difficulties in constructing accurate building models and handling uncertain disturbances. An adaptive robust model predictive control (ARMPC) is proposed to refine building models and handle uncertainty of disturbances. A model adaptation function is incorporated to perform online estimation of uncertain parameters of the building model using online measured building operation data, as the MPC controller is in operation. An additive uncertainty model to represent uncertainties of disturbances is integrated with the building model for robust optimization. The control performance of the ARMPC is compared with MPC controllers without adaptive modelling and robust optimization, as well as a conventional thermostat through simulation constructed based on a test building. When an energy-saving-biased setting is applied, ARMPC achieves the best thermal comfort performance among the tested controllers. The energy savings achieved by the ARMPC vary from ≈20% to ≈15%, compared to the thermostat, as uncertainty level of internal load increases from 0% to 60%. MPC controllers without adaptive modelling and robust optimization maintain ≈20% energy savings as the uncertainty level increases but at the expense of compromising thermal comfort. When a thermal-comfort-biased setting is applied, the MPC controllers maintain the indoor predicted mean vote (PMV) within a narrow range around thermal neutrality while achieving energy savings of around 10%, compared to the thermostat. The adaptive modelling and robust optimization of the ARMPC prevent the indoor condition from violating the constrains due to model inaccuracy and uncertainties in measured disturbances.

The effect of annealing on the self-organized morphology and component gradient distribution of films prepared from bimodal latexes blend containing 1:1 silicon-containing acrylate copolymer/silicon-free acrylate copolymer blend was studied using attenuated total reflectance–Fourier transform infrared (ATR-FTIR) spectroscopy, scanning electron microscopy with X-ray energy-dispersive (SEM-EDX) spectrometry, and atomic force microscopy (AFM). The distribution of silicon through the whole thickness of the film as a function of annealing was investigated using confocal Raman spectroscopy (CRS). AFM results show that poly(methyl methacrylate-co-n-butyl acrylate) latex fuses to form a continuous film at 25 °C. The wettability of the acrylate components and the heterogeneous composition of poly(3-[tris(trimethylsilyloxy)silyl] propyl methacrylate-co-methyl methacrylate) result in a graded block film. ATR-FTIR and SEM-EDX measurements reveal silicon-containing components segregate at the film–air interface upon annealing. CRS further shows that the nonlinear model gradient distribution of silicon is obtained, where the content of silicon component is enhanced and it gradually varies in the bulk. When the annealing temperature increases to 120 and 180 °C, blend latexes films demonstrate varying topography and phase images, indicating phase separation is induced by annealing. Furthermore, CRS implies that the destruction of the gradient structure is attributed to the phase separation of the two blend components.

Query auto-completion (QAC) is the first step of information retrieval, which helps users formulate the entire query after inputting only a few prefixes. Regarding the models of QAC, the traditional method ignores the contribution from the semantic relevance between queries. However, similar queries always express extremely similar search intention. In this paper, we propose a hybrid model FS-QAC based on query semantic similarity as well as the query frequency. We choose word2vec method to measure the semantic similarity between intended queries and pre-submitted queries. By combining both features, our experiments show that FS-QAC model improves the performance when predicting the user's query intention and helping formulate the right query. Our experimental results show that the optimal hybrid model contributes to a 7.54% improvement in terms of MRR against a state-of-the-art baseline using the public AOL query logs.