Search

The big data from various sensors installed on-board for monitoring the status of ship devices is very critical for improving the efficiency and safety of ship operations and reducing the cost of operation and maintenance. However, how to utilize these data is a key issue. The temperature change of the ship propulsion devices can often reflect whether the devices are faulty or not. Therefore, this paper aims to forecast the temperature of the ship propulsion devices by data-driven methods, where potential faults can be further identified automatically. The proposed forecasting process is composed of preprocessing, feature selection, and prediction, including an autoregressive distributed lag time series model (ARDL), stepwise regression (SR) model, neural network (NN) model, and deep neural network (DNN) model. Finally, the proposed forecasting process is applied on a naval ship, and the results show that the ARDL model has higher accuracy than the three other models. [ABSTRACT FROM AUTHOR]

How to explore the interaction between an individual researcher and others in scientific research, find out the degree of association among individual researchers, and evaluate the contribution of researchers to the whole according to the mechanism and law of interaction, is of great significance to grasp the overall trend of the field. Scholars mostly use bibliometrics to solve these problems and analyze the citation and cooperation among academic achievements from the dimension of "quantity". However, there is still no mature method for scholars to explore the evolution of knowledge and the relationship between authors; this paper tries to fill this gap. We narrow down the scope of research and focus the research content on the literature in biology and chemistry, collect all the papers from PubMed system (a very comprehensive authoritative database of biomedical papers) during 2014–2018, and take year as a specific analysis unit so as to improve the accuracy of the analysis. Then, we construct the author cooperation networks. Finally, through the above methods and steps, we identify the core authors of each year, analyze the recent cooperative relationships among authors, and predict some changes in the cooperative relationship among the authors based on the networks' analytical data, evaluating and estimating the role that authors play in the overall field. Therefore, we expect that the cooperative authorship networks supported by the complex network theory can better explain the author's cooperative relationship. [ABSTRACT FROM AUTHOR]

The recommendation algorithm in e-commerce systems is faced with the problem of high sparsity of users' score data and interest's shift, which greatly affects the performance of recommendation. Hence, a combined recommendation algorithm based on improved similarity and forgetting curve is proposed. Firstly, the Pearson similarity is improved by a wide range of weighted factors to enhance the quality of Pearson similarity for high sparse data. Secondly, the Ebbinghaus forgetting curve is introduced to track a user's interest shift. User score is weighted according to the residual memory of forgetting function. Users' interest changing with time is tracked by scoring, which increases both accuracy of recommendation algorithm and users' satisfaction. The two algorithms are then combined together. Finally, the MovieLens dataset is employed to evaluate different algorithms and results show that the proposed algorithm decreases mean absolute error (MAE) by 12.2%, average coverage 1.41%, and increases average precision by 10.52%, respectively. [ABSTRACT FROM AUTHOR]