LinkingPark: An automatic semantic table interpretation system

In this paper, we present LinkingPark, an automatic semantic annotation system for tabular data to knowledge graph matching. LinkingPark is designed as a modular framework which can handle Cell-Entity Annotation (CEA), Column-Type Annotation (CTA), and Columns-Property Annotation (CPA) altogether. It is built upon our previous SemTab 2020 system, which won the 2nd prize among 28 different teams after four rounds of evaluations. Moreover, the system is unsupervised, stand-alone, and flexible for multilingual support. Its backend offers an efficient RESTful API for programmatic access, as well as an Excel Add-in for ease of use. Users can interact with LinkingPark in near real-time, further demonstrating its efficiency.     

Determining robust reaction kinetics from limited data

Accurate chemical kinetics are essential for reactor design and operation. However, despite recent advances in “big data” approaches, availability of kinetic data is often limited in industrial practice. Herein, we present a comparative proof-of-concept study for kinetic parameter estimation from limited data. Cross-validation (CV) is implemented to nonlinear least-squares (LS) fitting and evaluated against Markov chain Monte Carlo (MCMC) and genetic algorithm (GA) routines using synthetic data generated from a simple model reaction. As expected, conventional LS is fastest but least accurate in predicting true kinetics. MCMC and GA are effective for larger data sets but tend to overfit to noise for limited data. LS-CV strongly outperforms these methods at much reduced computational cost, especially for significant noise. Our findings suggest that implementation of CV with conventional regression provides an efficient approach to kinetic parameter estimation with high accuracy, robustness against noise, and only minimal increase in complexity.     

Machine Learning-Based Two-Stage Data Selection Scheme for Long-Term Influenza Forecasting

One popular strategy to reduce the enormous number of illnesses and deaths from a seasonal influenza pandemic is to obtain the influenza vaccine on time. Usually, vaccine production preparation must be done at least six months in advance, and accurate long-term influenza forecasting is essential for this. Although diverse machine learning models have been proposed for influenza forecasting, they focus on short-term forecasting, and their performance is too dependent on input variables. For a country’s long-term influenza forecasting, typical surveillance data are known to be more effective than diverse external data on the Internet. We propose a two-stage data selection scheme for worldwide surveillance data to construct a long-term forecasting model for influenza in the target country. In the first stage, using a simple forecasting model based on the country’s surveillance data, we measured the change in performance by adding surveillance data from other countries, shifted by up to 52 weeks. In the second stage, for each set of surveillance data sorted by accuracy, we incrementally added data as input if the data have a positive effect on the performance of the forecasting model in the first stage. Using the selected surveillance data, we trained a new long-term forecasting model for influenza and perform influenza forecasting for the target country. We conducted extensive experiments using six machine learning models for the three target countries to verify the effectiveness of the proposed method. We report some of the results.     

Impact of air contamination by NOx on the performance of high temperature PEM fuel cells

High temperature PEM fuel cells show enhanced tolerances regarding fuel impurities like CO for use in various applications. However, the impact of air impurities like NO_x on the cell behavior is not completely understood yet. This study provides systematic investigation during 500 h of operation in presence of cathode air containing 10 ppm NO or NO₂. Nitrogen oxides provoke a strongly and linearly decreasing voltage of 245.3 ± 18.5 μV h⁻¹ and highly comparable damage that verifies similar HT-PEMFC degradation via both oxides. Cyclic voltammetry and electron microscopy reveal the loss of electrochemical catalyst surface by selectively poisoned surface and enforced catalyst particle growth. Impedance spectroscopy reveals besides increased electrode charge transfer resistances an affected proton conductivity. In contrast, SO₂/NO₂ impurity mixture in real occurring ratio causes less voltage decay due to a positive SO₂ impact through H₂SO₄ formation causing further shown and discussed effects like nitrate formation and discharge.     

Compression properties of gas diffusion layers and its constitutive model under cyclic loading

The compressive deformation of gas diffusion layer (GDL), which is highly nonlinear and related to the loading history, affects the performance of PEM fuel cell stacks. However, linear elastic models are widely used. In this study, a new nonlinear constitutive model is proposed to describe the compression properties. Macroscopic studies reveal that GDL has different mechanical properties during the first and repeated compression stages. Besides, the tangent modulus has a significant linear relationship with stress. The constitutive model can be rebuilt using the micro-mechanical theory of fiber assemblies by considering the bending of carbon fibers. Furthermore, a prediction method is proposed to describe cyclic compression behavior. The prediction results fit well with the test results with an average and maximum relative error of less than 5.30% and 18.13%, respectively. These conclusions are beneficial to the design of GDL with specific mechanical properties and the real-time analysis of PEM fuel cell.     

Structural and operational management of Turkish airports: a bootstrap data envelopment analysis of efficiency

This paper explores the structural and operational dimensions of the efficiencies of airports. The two-stage procedure is suggested to assess the efficiencies of airports in this study. In the first-stage, Classification and Regression Tree, which is one of the machine-learning approaches used to divide the airports into homogeneous and thus comparable sub-groups. In the second stage, the bootstrap data envelopment analysis approach obtains more precise structural and operational efficiency scores. To illustrate the proposed framework use, we applied it to a real case associated with Turkish airports. The results demonstrate that this framework presents a more comprehensive assessment of airport performance rather than conventional data envelopment analysis models. Moreover, it provides to show the deficiencies of the structural and operational management of airports. The findings can help anywhere airport authorities as well as Turkish airport authorities.     

Stream-Based Data Sampling Mechanism for Process Object

Process object is the instance of process. Vertexes and edges are in the graph of process object. There are different types of the object itself and the associations between object. For the large-scale data, there are many changes reflected. Recently, how to find appropriate real-time data for process object becomes a hot research topic. Data sampling is a kind of finding c hanges o f p rocess o bjects. There i s r equirements f or s ampling to be adaptive to underlying distribution of data stream. In this paper, we have proposed a adaptive data sampling mechanism to find a ppropriate d ata t o m odeling. F irst o f all, we use concept drift to make the partition of the life cycle of process object. Then, entity community detection is proposed to find changes. Finally, we propose stream-based real-time optimization of data sampling. Contributions of this paper are concept drift, community detection, and stream-based real-time computing. Experiments show the effectiveness and feasibility of our proposed adaptive data sampling mechanism for process object.     

Thermal and pH dual‐responsive hydrogels based on semi‐interpenetrating network of poly(N‐isopropylacrylamide) and collagen nanofibrils

Hydrogels with environment‐sensitive properties have great potential applications in the controlled drug release field. In this paper, hybrid hydrogels with semi‐interpenetrating polymer networks (semi‐IPNs), composed of poly(N‐isopropylacrylamide) (PNIPAM) as the thermo‐sensitive component by in situ polymerization and self‐assembled collagen nanofibrils as the pH‐sensitive framework, were prepared for controlled release of methyl violet as a model drug. From Fourier transform infrared spectroscopy and scanning electron microscopy, it was indicated that the crosslinking of PNIPAM in the presence of collagen nanofibrils led to the formation of semi‐IPNs with homogeneous porous structure, and the semi‐IPNs showed improved thermal stability and elastic properties compared with the native collagen as determined using differential scanning calorimetry and rheologic measurements. Furthermore, the semi‐IPNs possessed swelling behaviors quite different from those of neat collagen or PNIPAM hydrogel under various pH values and temperatures. Correspondingly, as expected, the drug release behavior in vitro for semi‐IPNs performed variously compared with that for single‐component semi‐IPNs, which revealed the tunable performance of semi‐IPNs for release ability. Finally the thermo‐ and pH‐responsive mechanism of the semi‐IPNs was illuminated to provide guidance for the application of the thermo‐ and pH‐sensitive collagen‐based hybrid hydrogels in controlled drug delivery systems. © 2019 Society of Chemical Industry     

基于间隔理论的过采样集成算法

针对传统集成算法不适用于不平衡数据分类的问题，提出基于间隔理论的AdaBoost算法（MOSBoost）。首先通过预训练得到原始样本的间隔；然后依据间隔排序对少类样本进行启发式复制，从而形成新的平衡样本集；最后将平衡样本集输入AdaBoost算法进行训练以得到最终集成分类器。在UCI数据集上进行测试实验，利用F-measure和G-mean两个准则对MOSBoost、AdaBoost、随机过采样AdaBoost（ROSBoost）和随机降采样AdaBoost（RDSBoost）四种算法进行评价。实验结果表明，MOSBoost算法分类性能优于其他三种算法，其中，相对于AdaBoost算法，MOSBoost算法在F-measure和G-mean准则下分别提升了8.4%和6.2%。