No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Abstract
In information network, different sources publish facts with different degrees of credibility and accuracy. To predict the truth values of the facts, several fact-finder algorithms are suggested which iteratively compute the trustworthiness of an information source and the accuracy of the facts it provides. However they ignore a great deal of relevant background and contextual information. In this paper, we proposed a novel maximum entropy weighted method to processing trust analysis, allowing us to elegantly incorporate knowledge such as the at tributes of the objects and the implications of the sources. Experiments demonst rate that our algorithm ignificantly improves performance over existing. ©, 2015, Chinese Institute of Electronics. All right reserved.
... Sen Hou et al. [18] proposed a maximum entropyweighted trust analysis algorithm based on information source clustering. The algorithm combines information such as object attributes and information source correlation, which effectively improves the performance of trust analysis. ...
... To evaluate the performance of the new model, several sets of simulation experiments were designed with reference to the trust model in Refs. [16,18,28], the experimental results were compared and analyzed. 1) Simulation environment and experimental data Simulation environment: The operating system is Windows 8; the compiler software is Matlab 7.0.1; the processor is the Intel (R) Core (TM) i7-6700 CPU @ 3.40GHz; and the internal memory is 32G. ...
... To advance the experiment, the window length is set to ten, the data set size is set to 50,000, and the new trust model is compared with the trust models in Refs. [16,18,28], through comparison of the simulated experiments; the results are plotted in the histogram shown in Fig.6. Fig.6 reveals that the precision and recall of the trust model presented in this paper are higher than the precision and recall of the other three trust models. ...
Due to the challenge of increasing data volume, the traditional trust model is unable to manage data with high efficiency and effectively extract useful information hidden in big data. To fully utilize big data and combine machine learning with trust evaluation, a trust evaluation model based on Long short‐term memory (LSTM) is presented. The powerful learning ability, expressive ability and dynamic timing of LSTM can be applied to study data while avoiding the vanishing and exploding gradient phenomena of traditional Recurrent neural networks (RNNs) to ensure that the model can learn sequences of random length and provide accurate trust evaluation. Targeting the performance instability caused by the LSTM model's random initialization of weights and thresholds, Particle swarm optimization (PSO), one of the intelligent algorithms, is introduced to find global optimal initial weights and thresholds. Experiments proved that the trust model proposed in this paper has high accuracy and contributes a new idea for trust evaluation in big data environments.
... Data acquisition and motor skills achievement summary of practitioner' motor skills related information in the early stage provide data support for practitioner' next training. Therefore, it is of positive significance to study the decision support system of motor skills evaluation in improving the scientificity and rationality of motor skills training [1,2]. ...
... The dynamic difference feature classification is used for information fusion, and the time-frequency feature decomposition of the data of motor skills decision support information is carried out. The transformation process is described as [9]: (2) In the formula, f (t) is the non-stationary instantaneous sampling value of characteristic data, p (a, b) is the joint distribution of mutual information, a is the cascade matching parameter, and b is the decomposition spectrum of windowed time-frequency domain. ...
In order to improve the performance of motor skills evaluation decision-making, data support is provided for the system through motor skills association data mining, optimum design of motor skills evaluation decision-making support system is carried out, the overall structure model of motor skills evaluation decision-making support system is constructed, the functional module framework and technical index analysis of the system are carried out, software development is carried out in the kernel structure of embedded Linux, and VIX bus is adopted. Technology is used to collect motor skills data. Design data mining technology based on mutual information feature extraction. Data mining code is loaded in the program loading module, and t/Embedded 4.6 is used to create the user interface of motor skills evaluation decision support system to realize visual control and software development design of the system. The simulation results show that the system has high accuracy and good reliability in mining motor skills evaluation decision-making information.
... On the basis of the reconstruction of the phase space of Library and information, the feature quantity of the semantic concept set of books and information is extracted [8][9][10] . The fuzzy clustering method is used to carry out the classified storage and information retrieval of the books and information, and the statistic of the library and information fusion is: X X X X (7) Adaptive learning is taken in the reconstructed phase space, the embedded dimension of the phase space is increased from m to 1 m + , and the semantic feature concept set of library and information fusion is obtained as follows: ...
In order to improve the ability of library and information management in colleges and universities, and improve intelligent retrieval level of books, a design method of library information management system is proposed based on big data fusion. The phase space reconstruction technology is used to reconstruct the feature of library and information. The feature quantity of semantic concept set of library information is extracted, and the classification storage and information retrieval of library information is carried out by fuzzy clustering method. The adaptive training method is used for feature fusion, and big data fusion of library and information is realized in high dimensional feature space. The data processing center is set up under the Linux kernel environment, the application program of the university library information management system is developed under the Linux kernel, and the VXI bus technology is used to transmit and schedule the university library information management information and data. Realize the software development and design of the school library information management system. The test results show that the design of university library information management system with this method has good information storage and scheduling ability, and it improves the performance of library information retrieval. In the information recall rate and recall rate and other indicators performance has an advantage.
Abstract—The performance of ad hoc networks depends on cooperation and trust among distributed nodes. To enhance security in ad hoc networks, it is important to evaluate trustworthiness of other nodes without centralized authorities. In this paper, we present an information theoretic framework to quantitatively measure trust and model trust propagation in ad hoc networks. In the proposed framework, trust is a measure of uncertainty with its value represented by entropy. We develop four Axioms that address the basic understanding of trust and the rules for trust propagation. Based on these Axioms, we present two trust models: entropy-based model and probability-based model, which satisfy all the Axioms. Techniques of trust establishment and trust update are presented to obtain trust values from observation. The proposed trust evaluation method and trust models are employed in ad hoc networks for secure ad hoc routing and malicious node detection. A distributed scheme is designed to acquire, maintain, and update trust records associated with the behaviors of nodes’ forwarding packets and the behaviors of making recommendations about other nodes. Simulations show that the proposed trust evaluation system can significantly improve the network throughput as well as effectively detect malicious behaviors in ad hoc networks.
The World Wide Web has become the most important information source for most of us. Unfortunately, there is no guarantee for the correctness of information on the Web. Moreover, different websites often provide conflicting information on a subject, such as different specifications for the same product. In this paper, we propose a new problem, called Veracity, i.e., conformity to truth, which studies how to find true facts from a large amount of conflicting information on many subjects that is provided by various websites. We design a general framework for the Veracity problem and invent an algorithm, called TRUTHFlNDER, which utilizes the relationships between websites and their information, i.e., a website is trustworthy if it provides many pieces of true information, and a piece of information is likely to be true if it is provided by many trustworthy websites. An iterative method is used to infer the trustworthiness of websites and the correctness of information from each other. Our experiments show that TRUTHFlNDER successfully finds true facts among conflicting information and identifies trustworthy websites better than the popular search engines.
This paper proposes a privacy preservation method based on random projection to overcome the curse of dimensionality in privacy preserving data mining. To prevent leaks of random matrix which can lead to the reconstruction attack, it first proposes the concepts of secure subspace and secure subspace mapping. Then, it constructs a secure subspace mapping using hash technique, which is implemented by a random projection matrix, and it achieves a low distortion embedding while preserving the data privacy. Finally, it proves that the secure subspace can preserve the Euclidean distance and inner product between any two original points. The experimental results show that the proposed technique can ensure the data quality in different data mining applications effectively under the precondition of preserving data privacy.
With the development of data mining technologies, privacy protection has become a challenge for data mining applications in many fields. To solve this problem, many privacy-preserving data mining methods have been proposed. One important type of such methods is based on Singular Value Decomposition (SVD). The SVD-based method provides perturbed data instead of original data, and users extract original data patterns from perturbed data. The original SVD-based method perturbs all samples to the same degree. However, in reality, different users have different requirements for privacy protection, and different samples are not equally important for data mining. Thus, it is better to perturb different samples to different degrees. This paper improves the SVD-based data perturbation method so that it can perturb different samples to different degrees. In addition, we propose a new privacy-preserving classification mining method using our improved SVD-based perturbation method and sample selection. The experimental results indicate that compared with the original SVD-based method, this new proposed method is more efficient in balancing data privacy and data utility.
Although much work in NLP has focused on simply determining what a document means, we also must know whether or not to believe it. Fact-finding algorithms attempt to identify the "truth" among competing claims in a corpus, but fail to take advantage of the user's prior knowledge and presume that truth itself is universal and objective rather than subjective. We introduce a framework for incorporating prior knowledge into any fact-finding algorithm, expressing both general "common-sense" reasoning and specific facts already known to the user as first-order logic and translating this into a tractable linear program. As our results show, this approach scales well to even large problems, both reducing error and allowing the system to determine truth respective to the user rather than the majority. Additionally, we introduce three new fact-finding algorithms capable of outperforming existing fact-finders in many of our experiments.