Search

Knowledge graph (KG) embedding projects the graph into a low-dimensional space and preserves the graph information. An essential part of a KG is the ontology, which always is organized as a taxonomy tree, depicting the type (or multiple types) of each entity and the hierarchical relationships among these types. The importance of considering the ontology during KG embedding lies in its ability to provide side-information, improving the downstream applications’ accuracy (e.g., link prediction, entity alignment or recommendation). However, the ontology has yet to receive adequate attention during the KG embedding, especially for instances where each entity may belong to multiple types. This ontology-enhanced KG embedding’s main challenges are twofold: determining how to discover the relationships among these types and how to integrate them with the entities’ relationship network. Although it is common to see attention-based models used in KG embedding, they cannot settle the issues raised simultaneously. Only a single type is assigned to each entity and the correlation among types are ignored in those models, leading to information loss and encumbered downstream tasks. To overcome these challenges, we propose a composite multi-type aware KG embedding model, whose main components are a multi-type layer and entity embedding layer. We model it as a natural language processing task at the multi-type layer to discover each entity’s multi-type feature and automatically capture their correlations. Additionally, a relation-based attention mechanism is conducted at the entity embedding layer, which aggregates neighborhoods’ information and integrates the multi-type layer’s information through common entities of these two layers. Through extensive experiments on two real KGs, we demonstrate that, compared to several state-of-the-art baselines, our Multi-Type aware Embedding (MTE) model achieves substantial gain in both Mean Rank and Hit@N for the link prediction task and accuracy for multi-type classification.

Solar thermochemical (STC) technology utilizes the entire spectrum of solar energy to decompose water to produce hydrogen. This technology reduces carbonic fuels, nearly only producing hydrogen rather than hydrogen-oxygen mixture. However, low water-splitting activity of redox materials restricts improvement of water-hydrogen conversion ratio and fuel production efficiency. Recently, a kind of perovskite LaFeO3 attracts attention, because of the good performance in photocatalysis hydrogen production. Nevertheless, how LaFeO3 system works in STC water-splitting cycle is rarely studied. In this paper, the first principle method at density functional theory level is adopted to reveal the hydrogen production mechanism of perovskite LaFeO3 doped with 25% Sr/Ca at A site. Hydrogen migration on material surface determines hydrogen generation rate. The activation energy of 25%-Ca-doped LaFeO3 is relatively lower 150.09 kJ/mol. In addition, fuel production efficiency has been calculated. When water to hydrogen conversion ratio is 100%, solar-to-fuel efficiency can reach maximum 0.472. The effect of water-splitting kinetics on hydrogen production is also discussed. The results indicate that when Tred = Toxi = T = 1200K and water to hydrogen conversion ratio is 10%, the dynamic efficiency of La0.75Ca0.25FeO3 can reach 20%. This research can provide index for improving the hydrogen production performance of STC technology.

In all areas of engineering, catastrophe assessment is an essential prerequisite for remedial action schemes. Modelers constantly push for more accurate models, and often meet goals by using increasingly complex, data mining-based blackbox models. However, system operators tend to favor interpretable models for after-the-fact preventive control (PC). While switching from blackbox to interpretable solutions, a tradeoff occurs between accuracy and interpretability. To avoid this tradeoff, we develop an intelligent framework for online catastrophe assessment and PC via a blackbox stacked denoising autoencoder (SDAE) equipped with accuracy and the ability to derive a PC scheme. Specifically, we implement a transient stability cost-sensitive assessment (TSCA) and PC case in the context of a power grid. First, using only controllable variables, we build the TSCA model by adding a sigmoid unit on top of the SDAE. Considering power systems’ conservatism, we explore a novel TSCA model’s training criterion to determine the operation conditions’ degrees of stability and divide them into three classes: stable, unstable, and boundary. Second, given an operation condition identified as unstable or boundary by TSCA model and its desired degree of stability, the PC model (the reverse of a TSCA model’s mapping) consists of the top sigmoid’s backward mapping and the stack of denoising decoders from trained SDAE. The former is formalized as an optimization problem to push back the desired degree of stability to a desired SDAE’s highest-level abstraction. The latter decodes back the desired SDAE’s highest-level abstraction to a desired operation condition (essentially, a PC scheme nearest to the controlled operation condition in the coordinates along the underlying causes that generate the observed data). This approach actually resembles operators’ tendency to adjust and stabilize unstable conditions (in terms of underlying causes) with the fewest control actions. A simulation study on the IEEE New England 39-bus system shows that, as a blackbox technology, our framework not only provides superior online situational awareness, but also finds a viable PC scheme, thereby justifying its practicability in engineering.

Owing to the wide applications of ammonia in hydrogen field and high energy consumption of the Haber-Bosch process, developing economic and environmentally benign ammonia synthesis process has attracted great interests. This work focuses on the moderately high two-step solar water-splitting of VN to produce ammonia, thus avoiding the reliance of fossil-fuel based heating source and pure hydrogen. Based on the equilibrium composition analysis, we find that V2O3, CH4 and N2 with mole ratio of 1:3:1.5 at TH = 1050 °C is enough for complete methane reforming and sufficient nitrogen fixation of V2O3. As for the water-splitting of VN, the production of NH3 is only possible at TL ≤ 400 °C, and inputting excessive water vapors is found to exert little effect on ammonia production at H2O:AlN>3:2. At the temperature range of full conversion between V2O3 and VN, the cycle efficiency, ηcycle, and solar-to-fuel efficiency, ηsolar-to-fuel, under different operating temperatures are compared, in which the highest ηcycle and ηsolar-to-fuel are 31.9% and 35.3% respectively. Moreover, efficiencies could be increased up to more than 37% with consideration of heat recuperation, demonstrating the great solar energy storage and fuel production potential of the proposed system.

For large-scale online social networks such as Facebook and Twitter, network analysis often uses graph queries to extract network information. Because of the work and memory required, usually such queries are performed in a distributed manner. However, most existing distributed graph computation systems optimize for offline graph analysis rather than online graph queries. The problem with this approach is that graph query tasks then must transfer a large volume of data and interactively answer queries within a short time frame. To resolve this, we propose a novel data-locality-aware task scheduling algorithm that optimizes interactive distributed graph queries. The scheduling algorithm jointly considers data placement and graph topology to reduce data transfer costs. After implementing the scheduling algorithm in a real-world distributed graph computation system, we evaluate the task scheduler’s effectiveness through simulations and real-life social graph queries. Results show that our scheduler reduces the querying time by one order of magnitude.

How to achieve fast and efficient resource allocation is an important optimization problem of resource management in cloud data center. On one hand, in order to ensure the user experience of resource requesting, the system has to achieve fast resource allocation to timely process resource requests; on the other hand, in order to ensure the efficiency of resource allocation, how to allocate multi-dimensional resource requests to servers needs to be optimized, such that server's resource utilization can be improved. However, most of existing approaches focus on finding out the mapping of each specific resource request to each specific server. This makes the complexity of resource allocation problem increases with the size of data center. Thus, these approaches cannot achieve fast and efficient resource allocation for large-scale data center. To address this problem, we propose a pattern based resource allocation mechanism based on the following findings. In a real-world cloud environment, the resource requests are usually classified into limited types. Thus, the mechanism first utilizes this feature to generate pattern information, which indicates which types of resource requests are suitable to be allocated together to a server. Then, the mechanism uses the pattern information as guidelines to make fast resource allocation decision and fully utilize server's multidimensional resources. Simulation experiments based on real and synthetic traces have shown that our mechanism significantly improves system's resource utilization and reduces the overall number of used servers.

A polyvinyl chloride (PVC) membrane sensor was developed to accurately detect beef taste quality grades. The beef taste was identified by the Fisher discriminant and verified by sensory evaluation. The PVC membrane sensor with the maximum open circuit potential (OCP) value and excellent performance was composed of 0.4 mL of dibutyl phthalate (DBP), 0.25 mL of lauryl ether phosphate, 3.6 mg of magnesium stearate and 6 mg of thermoplastic polyurethanes (TPU). Texture analyzer tests showed that the membranes showed desired physical properties, and the elasticity, viscosity and cohesion of the PVC membrane sensor were 4.74 mm, 2.17 mJ and 1.89 respectively. Scanning electron microscopy (SEM) showed the electrode immersed for 6 h without separation with PVC membrane. The coefficient of variation was less than 10% with a storage period of 7 months. Sensory evaluation showed that the beef sensory rated grades array determination by the Fisher discriminant method to verify the sensitivity of ammonia-sensitive substance-modified PVC membrane sensor had an accuracy of 93.3%.

The development of advanced materials relies on an innovative combination of two completely different materials. Loading quantum dots (QDs) into a three-dimensional (3D) network of hydrogels has proved to be a win-win strategy to enhance the synergy of the components. The resulting new QDs/hydrogel composites have a wide range of applications in important fields such as health, energy and environment. Herein, we review recent research progress of QDs/hydrogel composites. Synthetic strategies are classified into different categories according to the preparative sequence, a relatively more practical perspective, between the QDs and hydrogel. In addition, we summarize the latest applications of these nanocomposites (NCs) in the sensing field, with a focus on fluorescence sensing. We hope that this review can provide the reader with a comprehensive understanding on this composite system quickly so that they can choose the most suitable preparation method or sensing scheme based on actual conditions and requirements to obtain QDs/hydrogel composites with optimal performance.

Large-scale knowledge graphs containing millions of entities are very common nowadays. Querying knowledge graphs is essential for a wide range of emerging applications, e.g., question answering and semantic search. A star query aims to identify an entity by giving a set of related entities, which is an important query type on knowledge graphs. Answering star queries can be modeled as a graph query problem. Given a query graph Q , the graph query finds subgraphs in a knowledge graph G that match Q . We face two challenges on graph query: (1) existing graph query methods usually find subgraphs that are structurally similar to Q , which cannot measure whether a subgraph match satisfies the semantics of Q (i.e., real query intention), leading to an effectiveness issue, and (2) querying a large-scale knowledge graph is usually time-consuming because of the large search space. In this paper, we propose a Top- k semantic-aware graph query method over knowledge graphs for star queries, which provides semantically similar matches for Q instead of structurally similar matches. The semantic similarity of a match to Q is measured by an online computed bounding match score. By using bounds, we can efficiently prune the unpromising matches with lower semantic similarities without evaluating all matches. Extensive experiments over three real-world knowledge graphs confirm the effectiveness and efficiency of our solution.