Journal of Internet Technology最新文献

Chaos discrimination is a prerequisite for the application of chaos theory modeling. Since the average orbital period of an air traffic flow system is long, it is difficult to obtain time series with a small time scale and many data points, so the Small-Data Method is often adopted to quantitatively calculate the chaotic characteristic quantity. However, when using the power spectrum method, it is found that the Small-Data Method is prone to false judgments when the data volume is small. To reduce spurious judgments, we apply a chaos discrimination algorithm based on an Improved Alternative Data Method combined with the Small-Data Method for air traffic flow and analyze it by example. The algorithm was experimentally demonstrated to correct the false judgment results of the Small-Data Method. In particular, when the data volume is only 150, the discrimination accuracy of the improved algorithm is as high as 80%, which is 26% higher than the discrimination accuracy of the Small-Data Method. Moreover, the improved algorithm has better discriminative performance than the Small-Data Method under the same data volume condition, which is suitable for the chaotic discriminative analysis of the arrival traffic flow time series.

 

This work presents the creation and representation of an ontology model for the domain knowledge used for learning objects. The purpose of the developed ontology model is to define relations between learning objects that can be applied for their effective search and visualization. As the number of learning objects increases, the representation of the knowledge domain becomes challenging. In this paper, the authors propose the application of multidimensional concept maps (MCMs) for domain knowledge representation. The definition of different attributes used in the ontology model allow for defining the different dimensions needed for MCM ontology visualization. In order to achieve integration of the defined ontology model and MCMs, a software tool named Ontology-based system for learning objects retrieval (OBSLO) was developed. OBSLO’s role is to dynamically generate MCMs given the defined ontology with its relations and attributes, while also providing a content delivery environment and working space for learners. Proposed OBSLO architecture with integrated ontology model and MCMs was evaluated and compared to the learning management system where ontology and MCMs were not used. It was shown that learners using OBSLO showed better success rate in learning and positive level of satisfaction.

 

In the context of low-carbon city development, this paper further implements a rail transit passenger flow forecasting method to optimize energy consumption by combining the MMA allocation model with a two-tier planning model for carbon emission control. Through this approach, this paper not only fills the gap of rail transportation planning theories and methods compatible with low-carbon city development, but also emphasizes the importance of energy consumption in transportation planning. Based on a two-tier planning model, this paper considers the Starkberg game between multi-modal and multi-type passenger flow forecasting of rail transit and CO2 emissions of integrated transportation systems. By optimizing the allocation of users in the transportation network from the perspective of both users and planners, while optimizing the CO2 emissions of the integrated transportation system, the dual optimization of energy consumption and environmental benefits is achieved. The method will also be tested in Shanghai, and this paper will comparatively study three different carbon emission control schemes. By assigning passenger flows to the entire transportation system network in Shanghai based on information from the Fourth Integrated Transport Survey, including passenger flows on each road in the road network, passenger flows on each rail line, and characteristic indicators, this paper provides a reliable data base. This study provides a solid foundation for planning the layout of rail transit in a low-carbon mode and makes a positive contribution to sustainable urban development by optimizing energy consumption.

 

About 70% of the total cost of the water distribution system is used in the design of water distribution network (WDN), and selecting the most suitable pipe diameter for the WDN is the main way to reduce construction costs. The Rafflesia optimization algorithm (ROA) is a novel meta-heuristic algorithm, which was proposed recently. It has the characteristics of escaping local optimal solutions and stable performance. To further increase the solution quality and convergence speed of the algorithm, the opposition-based learning strategy is adopted in this paper to initialize the ROA algorithm population (namely the OBLROA algorithm). In this paper, the two-loop pipe network is taken as an actual test case, and the OBLROA algorithm is used to design the minimum cost pipe diameter combination. The experimental results show that the OBLROA algorithm can find the lowest cost pipe diameter combination of the two-loop pipe network under the constraints of pressure and velocity. Compared with some previous research work, the OBLROA algorithm needs the least number of evaluations to find the optimal solution, showing strong competitiveness.

 

To ensure the safe and stable operation of distance education systems, a security threat early warning technology based on blockchain is proposed for distance education system, which builds a security threat warning model. It uses the data interface in the interface layer to connect the teacher and student client. Then, the network behavior data of the distance education system is collected and transmitted to the data layer, where data blocks exchange the behavior data of the distance education system, and then the chain structured behavior data is generated and transmitted to the consensus layer. After the behavior data is transmitted to the incentive layer through the consensus layer, the distribution mechanism and basis are used to process and transfer the behavior data to the contract layer. The contract layer uses the threat early warning model to calculate the behavior data, and then conducts threat rating and early warning response on the data. It transmits the threat rating and early warning results to the application layer and presents them to users, thus realizing the security threat early warning of the distance education system. The experimental results show that the transcoding rate of this technology for the network behavior data of the distance education system is higher than 0.97, the early warning accuracy for the 10 types of network data of the distance education system can reach 100%, and the credibility of the early warning security threat of the types of DDOS IP, DDOS IP, phishing website URL address, and mobile malicious server IP address is higher than 0.96. Therefore, the technology has a strong capacity of behavior data storage in distance education systems, and can effectively warn different types of security threat in distance education systems. It has a more excellent application effect.

 

Large amounts of noise and a lack of contextual domain knowledge lead to slow and inefficient cross-domain image learning. This paper proposes an image scenario spatial data classification model based on evidence-based behavioral logic, intervenes in image annotation through evidence-based dynamic knowledge graphs, and uses spatial similarity measurement to evaluate the effectiveness and robustness of the method. The results show that: 1) Organizing the dynamic knowledge graphs of contextual domain knowledge by behavioral logic can significantly improve the association efficiency of each model. 2) The calculation method of image scenario space comparison based on behavior evidence can decrypt the implicit knowledge of images and significantly improve the effectiveness of image scenario space interpretation. The research results are helpful to guide the design and implementation of cross-domain image interpretation systems and improve the efficiency of information sharing.

 

Socio-meteorological observation is an essential part of meteorological information construction, where unofficial organizations and individuals (volunteers) are employed to collect meteorological data. Thanks to the participation of social forces, the density and richness of meteorological data are improved significantly, and hence more economical and social benefits are brought. However, problems such as privacy leakage and data islands hamper the sustainable development of socio-meteorological observation. To solve the problems, we propose a dynamic access control scheme with conditional anonymity in socio-meteorological observation. In the proposed scheme, conditional anonymity of volunteers is supported. On the one hand, the real identity of each valid volunteer is private; On the other hand, the real identity of the malicious volunteers will be revealed if they attempt to inject erroneous meteorological data into the system. In addition, a lazy update mechanism is designed, where the fluidity of the volunteers and attribute revocation of the data users are fully considered. Finally, we compare the proposed scheme with similar schemes theoretically and experimentally.

 

To tackle the uncertainties in life, a model that can efficiently convert qualitative concepts and quantitative values is essential. This model is referred to as a qualitative-quantitative uncertainty model. The conventional membership function provides a fixed membership degree that is incompatible with the fuzziness and randomness of qualitative concepts when a certain element of the theoretical domain is inputted. To address this issue, Academician Li introduced the cloud model, which is a qualitative-quantitative uncertainty model created for converting between qualitative and quantitative values. Unlike the traditional membership function, the cloud model generates a set of random numbers with a stable tendency that better captures the fuzziness and randomness of the qualitative concept when an element of the theoretical domain is inputted. In this paper, the background and fundamental concepts of cloud models are initially presented. Afterwards, we delve into the advancements of cloud models in various fields such as controller, data mining, and reliability. Through these discussions, the paper showcases the significant role that cloud models can play in resolving qualitative and quantitative conversion issues across different domains. The three numerical characteristics of cloud models are then described in detail, as well as cloud generator, virtual cloud and other cloud model related algorithms. Finally, some statistical properties of cloud models are discussed, as well as the current problems and future research directions.

 

This work presents a steganographic scheme based on Laplacian-of-Gaussian (LoG) edge detection and least significant bit (LSB) substitution. The cover image is first divided into continuous and non-overlapping 4×4-pixel blocks. The pixel at the top left corner of each block (first pixel) is defined as the reference pixel. After the LoG edge detection, the blocks are classified as edge or non-edge blocks, and this information is embedded in the reference pixels. Each non-reference pixel is then embedded with 5 bits and 4 bits of cipher text if it belongs to an edge block or non-edge block, respectively. Compared to the method of Tseng and Leng, Bai et al., and Ghosal et al., proposed method increased the capacity by 39.6%, 7.3%, and 42.7%, respectively, in the “Lena” cover image. To test the generalizability of our method, an embedding capacity and image quality test were conducted using 10,000 512 × 512 sized greyscale images from the BOSSBase dataset. Compared to the aforementioned previous methods, our method improved the capacity by 33.9%, 2.7%, and 36.1%, respectively, while maintaining an acceptable stego-image quality. Finally, proposed method can resist the detection of RS, pixel difference histogram analysis and second order SPAM features.

 

Expression recognition has wide application in the fields of distance education and clinical medicine. In response to the problems of insufficient feature extraction ability of expression recognition models in current research, and the deeper the depth of the model, the more serious the loss of useful information, a residual network model with multi-scale convolutional attention is proposed. This model mainly takes the residual network as the main body, adds normalization layer and channel attention mechanism, so as to extract useful image information at multiple scales, and incorporates the Inception module and channel attention module into the residual network to enhance the feature extraction ability of the model and to prevent the loss of more useful information due to too deep network, and to improve the generalization performance of the model. From results of lots of experiments we can see that the recognition accuracy of the model in FER+ and CK+ datasets reaches 87.80% and 99.32% respectively, with better recognition performance and robustness.