Search

Crowdsourcing is a hotspot research field which can facilitate machine learning by collecting labels to train models. Consequently, the state-of-the-art research efforts in crowdsourcing focus on truth inference or label integration, to remove inconsistent labels or to alleviate biased labeling. In turn, the integrated labels will be used to fine-tune machine learning models. Particularly, in this paper, we change the target of truth inference in crowdsourcing from discrete labels to multiple comments given by online participants, that is, the integration of the crowdsourced comments. For such a goal, we propose a Self-play and Sentiment-Emphasized Comment Integration Framework (SSECIF), based on deep Q-learning, with three unique features. First, our framework SSECIF can generate the comment integration in a totally self-play way, without relying on the ground truth generated by human effort. Second, the integrated comment generated by SSECIF can include salient content with low redundancy. Third, the proposed framework SSECIF has emphasized, with a higher intensity, the sentiment in the integrated comment, in order to reflect the attitude or opinion more obviously. Extensive evaluation on real-world datasets demonstrates that SSECIF has achieved the best overall performance in terms of both effectiveness and efficiency, compared with the state-of-the-art methods. Index Terms: Crowdsourcing; Comment Integration; Reinforcement Learning; Deep Q-Learning; Sentiment Analysis.

Nowadays, mining opinion or sentiment from the online user-generated texts has become a research hot-spot. Although, a large amount of lexicon-based Chinese polarity detection works have been done, yet the existing methods have one common flaw that even the same word can have opposite polarities among different seed-lexicons, or polarity fuzziness. Therefore, in order to further enhance the performance of Chinese sentiment polarity detection, in this paper, we start from a two-aspect lexicon expansion, specifically, detecting sentiment polarity for new words and revising sentiment polarity for words already defined in seed-lexicons, so that the polarity fuzziness can be avoided. Then, we formulate a novel sentiment polarity detection framework for Chinese (SPDFC) with more effort to fine-grained sentiment processing, involved in symmetrical mapping, sentiment feature pruning and text representation. In this way, the word polarity can be directly taken as features, further penetrating into polarity detection phase. According to the experimental results, compared to other classical and state-of-the-art methods, the proposed the framework SPDFC can achieve the best overall performance from the perspective of Chinese polarity detection, sentiment feature pruning as well as text representation.

Clustering is an important research direction in data mining. However, there is no one clustering algorithm that can be applied efficiently in all situation. Clustering ensemble is the best way to solve the above-mentioned problems. It combines the results of multiple clustering algorithms, and the final result is significantly better than a single clustering algorithm. Although there is a lot of constraint information, the existing clustering ensemble algorithm does not utilize it. This paper uses constraint information in consensus function and proposes a Semi-supervised Selective Clustering Ensemble based on Chameleon (SSCEC) and Semi-supervised Selective Clustering Ensemble based on Ncut (SSCEN) to solve the above problem. SSCEC uses the chameleon algorithm as consensus function, and processes constraint information in subgraph partition and subgraph combining. SSCEN uses the Normalized cut algorithm as consensus function, and processes constraint information in the process of graph dichotomy. The experiment results show that our proposed two semi-supervised member selection clustering ensemble algorithms are better than other semi-supervised algorithms.

The typical aim of user matching is to detect the same individuals cross different social networks. The existing efforts in this field usually focus on the users' attributes and network embedding, but these methods often ignore the closeness between the users and their friends. To this end, we present a friend closeness based user matching algorithm (FCUM). It is a semi-supervised and end-to-end cross networks user matching algorithm. Attention mechanism is used to quantify the closeness between users and their friends. We considers both individual similarity and their close friends similarity by jointly optimize them in a single objective function. Quantification of close friends improves the generalization ability of the FCUM. Due to the expensive costs of labeling new match users for training FCUM, we also design a bi-directional matching strategy. Experiments on real datasets illustrate that FCUM outperforms other state-of-the-art methods that only consider the individual similarity.

The combination of Unmanned Aerial Vehicle (UAV) technologies and computer vision makes UAV applications more and more popular. Computer vision tasks based on deep learning usually require a large amount of task-related data to train algorithms for specific tasks. Since the commonly used datasets are not designed for specific scenarios, in order to give UAVs stronger computer vision capabilities, large enough aerial image datasets are needed to be collected to meet the training requirements. In this paper, we take low-altitude aerial image object detection as an example to propose a framework to demonstrate how to construct datasets for specific tasks. Firstly, we introduce the existing low-altitude aerial images datasets and analyze the characteristics of low-altitude aerial images. On this basis, we put forward some suggestions on data collection of low-altitude aerial images. Then, we recommend several commonly used image annotation tools and crowdsourcing platforms for data annotation to generate labeled data for model training. In addition, in order to make up the shortage of data, we introduce data augmentation techniques, including traditional data augmentation and data augmentation based on oversampling and generative adversarial networks.

Conversation generation is an important natural language processing task and has attracted much attention in recent years. The realization of the conversation model is also of great significance to the field of social computing, helping to build artificial intelligence robots on social networks. The open domain conversation model is fundamentally data-driven, which can be roughly divided into retrieval models and generation models. Although remarkable progress has been achieved in recent years, it is still difficult to get responses that are grammatically and semantically appropriate. We propose the Rerank of Retrieval-based and Transformer-based Conversation model (RRT), a novel conversation model that combines the retrieval model with the generation model for the purpose of obtaining context–appropriate response. The context–response pairs with the highest similarity from training set are retrieved using traditional retrieval method, and further ranked to obtain the retrieval candidate response. We replaced the traditional sequence-to-sequence models for conversation generation by the transformer model and achieved better results with less training time. Finally, the post-reranking module is used to rank the retrieved candidate and the generated one to obtain the final response. We conducted detailed experiments on two datasets and the results show that compared with the traditional generation model, our model has a significant improvement in each metric, and the training time is decreased by a factor of 5. Furthermore, our model is more informative and relevant to the input context than the retrieval model.

In this article, first a comprehensive study of the impact of term weighting schemes on the topic modeling performance (i.e., LDA and DMM) on Arabic long and short texts is presented. We investigate six term weighting methods including Word count method (standard topic models), TFIDF, PMI, BDC, CLPB, and CEW. Moreover, we propose a novel combination term weighting scheme, namely, CmTLB. We utilize the mTFIDF that takes into account the missing terms and the number of the documents in which the term appears when calculating the term weight. For further robust term weight, we combine mTFIDF with two weighting methods. We evaluate CmTLB against the studied weighting schemes by the quality of the learned topics (topic visualization and topic coherence), classification, and clustering tasks. We applied weighting schemes to Latent Dirichlet allocation (LDA) and Dirichlet multinomial mixture (DMM) on eight Arabic long and short document datasets, respectively. The experiment results outline that appropriate weighting schemes can effectively improve topic modeling performance on Arabic texts. More importantly, our proposed CmTLB significantly outperforms the other weighting schemes. Secondly, we investigate whether the Arabic stemming process can improve topic modeling performance. We study the three approaches of Arabic stemming including root-based, stem-based, and statistical approaches. We also train topic models with weighting schemes on documents after applying four stemmers related to different stemming approaches. The results outline that applying the stemming process not only reduces the dimensionality of term-document matrix leading to fast estimation process, but also show enhancement of topic modeling performance both on short and long Arabic documents. Moreover, Farasa stemmer achieves the highest performance in most cases, since it prevents the ambiguity that may happen because of the blind removal of the affixes such as in root-based or stem-based stemmers.

Most current automatic summarization methods are for English texts. The distinction between words in Chinese text is large, the types of parts of speech are many and complex, and polysemy or ambiguous words appear frequently. Therefore, compared with English text, Chinese text is more difficult to extract useful feature words. Due to the complex syntax of Chinese, there are currently relatively few automatic summarization methods for Chinese text. In the past, only the important sentences in the original text can be selected and simply arranged to obtain a summary with chaotic sentences and insufficient coherence. Meanwhile, because Chinese short text usually contains more redundant information and the sentence structure is not neat, we propose a topic-based automatic summary method for Chinese short text. Firstly, a key sentence selection method is proposed combining topic words and TF-IDF to obtain the score of each text corresponding to the topic in the original text data. Then the sentence with the highest score as the topic sentence of the topic is selected. Considering that the short text of Weibo may contain a lot of irrelevant information and sometimes even lack some important components of topic, three retouching mechanisms are proposed to improve the conciseness, richness and readability of topic sentence extraction results. We validate our approach on natural disaster and social hot event datasets from Sina Weibo. The experimental results show that the polished topic summary not only reflects the exact relationship between topic sentences and natural disasters or social hot events, but also has rich semantic information. More importantly, we can almost grasp the basic elements of natural disaster or social hot event from the topic sentence, so as to help the government guide disaster relief or meet the needs of users for quickly obtaining information of social hot events.

Nowadays, the amount of user-generated contents (UGCs) or texts has surged exponentially. Therefore, recognizing emotions from these texts can bring about lots of advantages. In this paper, we have proposed a novel model named Deep Rolling to predict emotion for target participant in a multi-participant communication context. First, the proposed method converts a text collection into a set of n -dimension vectors for emotion representation and re-organizes texts into a sequence in time order. Then, Deep Rolling can predict the emotion of target participant corresponding to a future time point . Second, apart from simply taking in texts posted by target participant via LSTM , the proposed method has also incorporated texts posted by other participants at every time step by CNN . In this way, Deep Rolling can predict target participant’s emotion by processing emotions from both the target and all the other participants in an ensemble way. Finally, data factorization has also been introduced into Deep Rolling to enhance the overall prediction efficiency. According to experimental results, compared with the state-of-art methods, our proposed model has achieved the best prediction precision on different target participants. At the same time, Deep Rolling has also maintained the prediction efficiency at an acceptable level.

Community detection is a classic and essential task in complex network analysis which aims at understanding their structural properties and functional organization. Community detection based on overlapping nodes has become one of the most popular methods in recent years. But, how to define the overlapping nodes in the real network is an important job. In this paper, we use the speaker-listener label propagation algorithm to find overlapping nodes. Second, we proposed a new metric which is based on local and global attributes to measure the node influence so as to evaluate the overlapping nodes’ importance. Community detection is realized with the $k$ -lowest-influence overlapping nodes deleting. Then, we assign the removed nodes into specific communities by voting to find the final community structure. The voting strategy is based on choosing the specific communities containing most of their neighbors. Finally, the extensive experiments on real-world networks demonstrate that our proposed method improves the quality of community detection methods and show both the effectiveness and efficiency of the method.

The natural characteristics of Location Based Services (LBS) cause potential threats to location privacy. Users need to send their current locations to get the service, which may lead to the leakage of their location privacy. An effective way to protect user’s location privacy is to use an imprecise location and send this region to the Cloud-of-things system to replace his real location. When user moves and sends continuous queries, the user needs to transmit the region to the server continuously and an attacker is able to infer the real location from the overlapping regions. In this paper, we propose a novel location privacy pre-protection method for cloud-of things system to preserve user’s trajectory privacy. User’s moving behaviors are analyzed through Mobility Markov chain. The proposed location cloaking algorithm enlarges the small area to satisfy the user’s privacy requirements, so that the location trajectory privacy in the environment of cloud-of-things system is addressed efficiently. Experimental results show the performance of our method in terms of the number of moving steps, the cloaking threshold value, in addition to the user chosen anonymity value.

Accurately labeling training data plays a critical role in various supervised learning tasks. Now a wide range of algorithms have been developed to identify and remove mislabeled data as labeling in practical applications might be erroneous due to various reasons. In essence, these algorithms adopt the strategy of one-zero sampling (OSAM), wherein a sample will be selected and retained only if it is recognized as clean. There are two types of errors in OSAM: identifying a clean sample as mislabeled and discarding it, or identifying a mislabeled sample as clean and retaining it. These errors could lead to poor classification performance. To improve classification accuracy, this paper proposes a novel probabilistic sampling (PSAM) scheme. In PSAM, a cleaner sample has more chance to be selected. The degree of cleanliness is measured by the confidence on the label. To accurately estimate the confidence value, a probabilistic multiple voting idea is proposed which is able to assign a high confidence value to a clean sample and a low confidence value to a mislabeled sample. Finally, we demonstrate that PSAM could effectively improve the classification accuracy over existing OSAM methods.

With the development of mobile Internet technology, location-based services (LBS) has been widely used in social networks, such as WeChat. A user of WeChat can discover people nearby with the help of LBS. However, users need to send their current locations to get the service, which may lead to the leakage of their location privacy. An effective way to protect the user's location privacy is using an imprecise location, such as a region to replace his real location. When processing continuous queries, the user need to transmit the region to the server continuously and an attacker is able to infer the real location from the overlapping regions. In this paper, we use circle as the query region and describe how to form a circle region. Then we study the overlapping circle attack and design a method to avoid this attack by changing the confidence level to a fixed value. We also propose an algorithm E PPINQ for the server to process a k NN query while protecting user's location against the attack. Experimental results show that our algorithm is more effective and efficient than the existing algorithm.

Cache memory plays an important role in improving the performance of web servers, especially for big data transmission, which response time is constrained. It is necessary to use an effective method, such as web cache. Because an outstanding cache replacement algorithm can not only reduce the users access time but also improve the performance of the system. The traditional used weighting replacement policy does not consider the size parameter, hence, it may perform poorly while the datasets are larger. In this paper, we propose a novel, high-performance cache replacement algorithm for the web cache, named weighting size and cost replacement policy (WSCRP) bases on the weighting replacement policy. The algorithm recalculates the objects weight with adding the cost attribute in the cache, then orders the weight. Additionally the influence of various factors on the Web object as frequency, time, and cost value are considered. When the cache space cannot satisfy the new request object, the replacement policy WSCRP replaces the largest weighting and cost object. The experiments show that proposed algorithm has higher hit rate and byte rate for different datasets, and can effectively improve the performance of web cache.

Covert communication is an extension of the traditional encryption technology, which hides information through carriers such as text and images. Traditional covert communication relies on a third-party central node, which is easy to be detected and attacked. As a new type of decentralized technology, blockchain has been widely developed in recent years. Its anti-interference, anti-tampering and other characteristics can effectively alleviate the problems faced by traditional covert communications. Combining with blockchain technology to design an effective and feasible hidden communication model and method becomes the focus of research. To this end, this paper designs a new covert communication model suitable for Ethereum based on the whisper protocol. First, we introduce the covert communication model under the blockchain, as well as the whisper protocol in Ethereum. Then, based on the characteristics and structure of the whisper protocol, a new type of covert communication method was designed. Finally, the anti-jamming, anti-tampering and anti-detection of the model are proved, and its information embedding and transmission efficiency are also evaluated by comparing with the time-based covert communication.

The large-scale and complex structure of real networks brings enormous challenges to traditional community detection methods. In order to detect community structure in large-scale networks more accurately and efficiently, we propose a community detection algorithm based on the network embedding representation method. Firstly, in order to solve the scarce problem of network data, this paper uses the DeepWalk model to embed a high-dimensional network into low-dimensional space with topology information. Then, low-dimensional data are processed, with each node treated as a sample and each dimension of the node as a feature. Finally, samples are fed into a Gaussian mixture model (GMM), and in order to automatically learn the number of communities, variational inference is introduced into GMM. Experimental results on the DBLP dataset show that the model method of this paper can more effectively discover the communities in large-scale networks. By further analyzing the excavated community structure, the organizational characteristics within the community are better revealed.

With the popularity of location-aware devices (e.g., smart phones), a large number of trajectory data were collected. The trajectory dataset can be used in many fields including traffic monitoring, market analysis, city management, etc. The collection and release of trajectory data will raise serious privacy concerns for users. If users’ privacy is not protected enough, they will refuse to share their trajectory data. In this paper, a new trajectory privacy protection method based on random sampling differential privacy (TPRSDP), which can provide more security protection, is proposed. Compared with other methods, it takes less time to run this method. Experiments are conducted on two real world datasets to validate the proposed scheme, and the results are compared with others in terms of running time and information loss. The performance of the scheme with different parameter values is verified. The setting of the new scheme parameters is discussed in detail, and some valuable suggestions are given.

As a kind of noise, mislabeled training data exist in many applications. Because of their negative effects on learning, many filter techniques have been proposed to identify and eliminate them. Ensemble learning-based filter (EnFilter) is the most widely used filter which employs ensemble classifiers. In EnFilter, first the noisy training dataset is divided into several subsets. Each noisy subset is then checked by the multiple classifiers which are trained based on other noisy subsets. It is noted that since the training data used to train multiple classifiers are noisy, the quality of these classifiers cannot be guaranteed, which might generate poor noise identification result. This problem is more serious when the noise ratio in the training dataset is high. To solve this problem, a straightforward but effective approach is proposed in this work. Instead of using noisy data to train the classifiers, nearly noise-free (NNF) data are used since they are supposed to train more reliable classifiers. To this end, a novel NNF data extraction approach is also proposed. Experimental results on a set of benchmark datasets illustrate the utility of our proposed approach.

Community detection in social networks is a fundamental task of complex network analysis. Community is usually regarded as a functional unit. Networks in real world more or less have overlapping community structure while traditional community detection algorithms assume that one vertex can only belong to one community. This paper proposes an efficient overlapping community detection algorithm named LED (Loop Edges Delete). LED algorithm is based on Structural Clustering, which converts structural similarity between vertices to weights of network. The evaluations of the LED algorithm are conducted both from classical networks from literature and C-DBLP, which is a huge and real-life co-author social network in China. The results show that LED is superior to other methods in accuracy, efficiency, comparing with FastModurity and GN algorithm.

Differential privacy can provide provable privacy security protection. In recent years, a great improvement has been made, however, in practical applications, the utility of original data is highly susceptible to noise, and thus, it limits its application and extension. To address the above problem, a new differential privacy method based on smooth sensitivity has been proposed in this paper. Using this method, the dataset's utility is improved greatly by reducing the amount of noise that is added, and this was validated by experiments.

Medical service providers offer their patients high quality services in return for their trust and satisfaction. The Internet of Things (IoT) in healthcare provides different solutions to enhance the patient-physician experience. Clinical Decision-Support Systems are used to improve the quality of health services by increasing the diagnosis pace and accuracy. Based on data mining techniques and historical medical records, a classification model is built to classify patients&rsquo, symptoms. In this paper, we propose a privacy-preserving clinical decision-support system based on our novel privacy-preserving single decision tree algorithm for diagnosing new symptoms without exposing patients&rsquo, data to different network attacks. A homomorphic encryption cipher is used to protect users&rsquo, data. In addition, the algorithm uses nonces to avoid one party from decrypting other parties&rsquo, data since they all will be using the same key pair. Our simulation results have shown that our novel algorithm have outperformed the Na&iuml, ve Bayes algorithm by 46.46%, in addition to the effects of the key value and size on the run time. Furthermore, our model is validated by proves, which meet the privacy requirements of the hospitals&rsquo, datasets, frequency of attribute values, and diagnosed symptoms.

As a research branch of data mining, clustering, as an unsupervised learning scheme, focuses on assigning objects in the dataset into several groups, called clusters, without any prior knowledge. Density-Based Spatial Clustering of Applications with Noise (DBSCAN) is one of the most widely used clustering algorithms for spatial datasets, which can detect any shapes of clusters and can automatically identify noise points. However, there are several troublesome limitations of DBSCAN: (1) the performance of the algorithm depends on two specified parameters, e and MinPts in which e represents the maximum radius of a neighborhood from the observing point and MinPts means the minimum number of data points contained in such a neighborhood. (2) The time consumption for searching the nearest neighbors of each object is intolerable in the cluster expansion. (3) Selecting different starting points results in quite different consequences. (4) DBSCAN is unable to identify adjacent clusters of various densities. In addition to these restrictions about DBSCAN mentioned above, the identification of border points is often ignored. In our paper, we successfully solve the above problems. Firstly, we improve the traditional locality sensitive hashing method to implement fast query of nearest neighbors. Secondly, several definitions are redefined on the basis of the influence space of each object, which takes the nearest neighbors and the reverse nearest neighbors into account. The influence space is proved to be sensitive to local density changes to successfully reduce the amount of parameters and identify adjacent clusters of different densities. Moreover, this new relationship based on the influence space makes the insensitivity to the ordering of inputting points possible. Finally, a new concept-core density reachable based on the influence space is put forward which aims to distinguish between border objects and noisy objects. Several experiments are performed which demonstrate that the performance of our proposed algorithm is better than the traditional DBSCAN algorithm and the improved algorithm IS-DBSCAN.

To solve the scheduling problem of workflow tasks in cloud computing, this paper combined the improved fuzzy c-means clustering algorithm (IFCM) and the improved ant colony optimization algorithm (IACO) and proposed a new workflow task scheduling algorithm. Firstly, the proposed algorithm used the IFCM to classify resources. Then, tasks will be sorted by their priority. Based on the results of resource clustering and the distance between resources and expect of tasks, tasks will be assigned to the appropriate resources and the scheduling will be initialized. After that, the workflow tasks will be encoded based on the initial scheduling. At last, ant colony optimization algorithm will be improved by the cross and mutation operation in genetic algorithm and used to search optimal schedules. The experiments showed that the proposed algorithm could quickly and efficiently find appropriate scheduling scheme, effectively reduce the time span of workflow tasks and increase the utilization of resources.

Chinese Digital Bibliography & Library Project C-DBLP is a huge and real-life co-author social network in China, rarely cited by published paper. It contains a large amount of ground-truth community structure with distinguished research topics. Despite the fact that rich studies on community detection have been conducted with gains of practically fruitful algorithms, unfortunately, with the coming of 'Big Data' era and speedy development of mobile devices, social networks like C-DBLP have incredibly expanded on nodes and edges, as a result, because of massive data cardinality, a large portion of community detection methods consume memory resource excessively. Therefore, in this work, we select Based on Structural Connection Hierarchical Exploration BSCHE algorithm to partition nodes in C-DBLP because of its On time cost, fast enough to process massive data, and its novel physical meaning of similarity between nodes defined by structural connection and availability. In addition, in order to avoid huge memory resource consumption caused by 'Big Data' of C-DBLP, we strengthen BSCHE as a framework BSCHEF by our proposed 'count-pointer-strategy' imitated from incremental batch process to detect co-author communities on C-DBLP. The experiment results show that BSCHEF can find sets of communities on C-DBLP more effectively with the highest modularity value and the least execution time compared to other clustering algorithm. Copyright © 2015 John Wiley & Sons, Ltd.

Classification techniques are important in bioinformatics analysis as they can separate various bioinformatical data into distinct groups. To obtain good classifiers, accurate labeling of the training data is required. However labeling in practical bioinformatics applications might be erroneous due to various reasons. To identify those mislabeled data, an ensemble learning based scheme, single-voting has been widely used. It generates multiple classifiers and makes use of their voting to detect mislabeled data. Single-voting scheme mainly consists of two components: data partitioning component to generate multiple classifiers, and mislabeled detection component to identify mislabeled data. Existing works in this field mainly focus on mislabeled detection part and neglect data partitioning. However, our analysis shows that data partitioning plays an important role in single-voting scheme. This analysis helps us proposing a novel multiple-voting scheme. It is superior to traditional single-voting by reducing the unreliable influence from data partitioning. Empirical and theoretical evaluations on a set of bioinformatics datasets illustrate the utility of our proposed scheme.

Nowadays, information sharing as an indispensable part appears in our vision, bringing about a mass of discussions about methods and techniques of privacy preserving data publishing which are regarded as strong guarantee to avoid information disclosure and protect individuals' privacy. Recent work focuses on propos ing different anonymity algorithms for varying data publishing scenarios to satisfy privacy requirements, and keep data utility at the same time. K-anonymity has been proposed for privacy preserving data publishing, which can prevent linkage attacks by the means of anonymity operation, such as generalization and suppression. Numerous anonymity algorithms have been utilized for achieving k-anonymity. This paper provides an overview of the development of privacy preserving data publishing, which is restricted to the scope of anonymity alg orithms using generalization and suppression. The privacy preserving models for attack is introduced at first. An overview of seve ral anonymity operations follow behind. The most important part is the coverage of anonymity algorithms and information metric which is essential ingredient of algorithms. The conclusion and perspective are proposed finally.

Server performance is one of the critical activities in the data grid environment. A large number of applications require access to huge volumes of data from grid servers. In this case, efficient, scalable and robust grid server which can deal with large file transfer concurrent is needed. In this paper, we analyze the bottleneck of our grid servers and introduce user-space I/O scheduling, zero copy and event-driven architecture in our grid server to improve the servers’ performance. The user-space I/O scheduling can save almost 50% I/O time in a huge number of small files transfer. Grid servers can elimination CPU consumptions between kernel and user space by zero copy and cut 63% times for context switches. Event-driven architecture will save 30% CPU usage to reach the best performance by thread-driven architecture. Optimization method combination of these three above are used in our grid servers, the full-load throughput of our system is 30% more than traditional solutions and only 60% CPU consumed compared with traditional solutions.

An improved fuzzy c-means algorithm is put forward and applied to deal with meteorological data on top of the traditional fuzzy c-means algorithm. The proposed algorithm improves the classical fuzzy c-means algorithm (FCM) by adopting a novel strategy for selecting the initial cluster centers, to solve the problem that the traditional fuzzy c-means (FCM) clustering algorithm has difficulty in selecting the initial cluster centers. Furthermore, this paper introduces the features and the mining process of the open source data mining platform WEKA, while it doesn’t implement the FCM algorithm. Considering this shortcoming of WEKA, we successfully implement the FCM algorithm and the advanced FCM algorithm taking advantage of the basic classes in WEKA. Finally, the experimental clustering results of meteorological data are given, which can exactly prove that our proposed algorithm will generate better clustering results than those of the K-Means algorithm and the traditional FCM algorithm.

As a research branch of grid computing, data grid focuses on the management of large-scale distributed data sets. Replica management is one of the most important issues in the data grid, which can offer fast data access time, high data availability and low bandwidth consumption. Computing Intelligent Algorithm (CIA) has been proved to be effective in the solution of large-scale distributed computing problems, whereas Quantum Evolutionary Algorithm (QEA) is one of these excellent optimization algorithms and little literatures are made for its application in Data Grid Replica Management (DGRM). This paper focuses on the application of the QEA in data grid replica creation strategy. A QEA-based global replica creation strategy is proposed after reviewing the replica creation strategies. The optimization model is divided into single and multi data replica creation two parts. The representation, evaluation and constraint procedure three key technologies problems for each part are discussed in detail. The detail algorithm of QEA based replica creation is provided. The experiments were carried out with OptorSim, and the results have shown that QEA-based replica creation strategy can effectively reduce the job response time and network bandwidth consumption, comparing to Genetic Algorithms (GAs), Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO) algorithms. Especially, its performance becomes better and better with the incensement of the number of jobs. The non-parametric statistical tests are used to verify the significant of QEA.

Meteorological observational data is collected from many stations around the country at the same time, and immediately submitted to the national meteorological data center-NMIC. NMIC aggregates national data and carries out effective processing and rapid service. This paper describes a high performance computing model in cluster environment to process massive meteorological data, which employs parallel computing and distributed computing frameworks focused on task division, task scheduling management, load balancing methods on various types of meteorological data in the Master-Slave operation environment. Within implementation of this architecture, high ability to efficiently handle large amount of content and good usability and stability of the data management and date service are achieved.

The ego network is a network of a user with his friends. The social network analysis method has provided some methods to help users classify their friends, including manually categorising friends or system classification. Whereas, categorising friends manually is time consuming. In this paper, we will discuss how to realise community identification automatically and accurately. To achieve this, we propose a method which utilises not only the similarity of user attributes but also the features of network structure and friends contact frequency. On the basis of the users profile, we identify the relationship between them firstly. Second, we realise community identification using the structure features. Third, we introduce contact frequency to identify the relationship between users and their friends more accurately. Extensive experiments on real-world data show that our approach outperforms the state-of-the-art technique, in terms of balance error rate and F1 score.

In this paper, we adopt the classification of personal information and hierarchy of services to build a privacy system, in which one communicates with each other via pipes with different security levels. In each level, one has the corresponding rights to access each other. The requesters are not able to be infringed based on the personal information that service obtains from service providers. Privacy system can decrease the interaction, while in other circumstance the system strengthens and enhances the privacy preserving. Thus we strike a balance between two goals of Ubiquitous Computing: interaction and privacy preserving.

Data classification has attracted many researchers’ attention. Many evolutionary algorithms (EAs) were employed to take advantage of their global search ability. In supervised classification research issues, EAs were only used to improve the performance of classifiers either by optimizing the parameters or structure of the classifiers, or by pre-processing the inputs of the classifiers. Although genetic programming or evolutionary based decision tree approaches are proposed for classification, the development of these approaches is limited by their special structure. In this paper, we propose a new mathematical optimization model for the supervised classification problem and use fireworks algorithm (FWA) to do classification directly without modification. In the new optimization model, a linear equations set is constructed based on training set, and we propose an objective function which can be optimized by FWA. Four different data sets have been employed in the experiments, and 70% samples are used as train sets while the rest are used as test sets. The results show that the label of test sets can be identified accurately by our new methods. This paper also shows that the optimization model can be used for classification and employing EA to solve this optimization model is feasible.

As one of the important issues of data classification, classification has attracted the attention of many researchers in the field of data mining. Different from clustering research issues, in classification research issues, evolutionary clustering algorithms (EAs) were only used to improve the performance of classifiers either by optimizing the parameters or structure of the classifiers, or by pre-processing the inputs of the classifiers. Lots of evolutionary algorithms are employed to solve unsupervised classification, i.e., clustering. In this article, we will create a new mathematical model for supervised classification problem and use brain storm optimization algorithm (BSO) to search the global optimal solution, which resolved the problem of supervised classification with new ideas. The main objective is to find a better method. By using a new classification algorithm based on BSO, we are looking forward to optimizing the result.

Classification and clustering are two important problems in machine learning and data mining. Recently, many evolutionary computation (EC) techniques have been employed to solve the clustering problems by taking the advantage of the global search ability of EC. Different from the situation in clustering research issues, EC techniques are only employed to improve the performance of the classifiers either by optimising their parameters and structures, or by pre-processing their inputs. In this paper, we propose an evolutionary optimisation classification model for the classification problems and the fireworks algorithm (FWA) is employed to solve the classification model directly. In the optimisation classification model, a linear equation set is constructed based on the training sets and an objective function which can be optimised by the EC techniques is proposed according to the equation set. Ten different datasets have been employed in the experiments. For each dataset, 70% instances are used as a training set while the rest are used as a test set. The experiment results show that it is possible to directly solve the classification problems by EC techniques through introducing the evolutionary optimisation classification model. Moreover, FWA performs better than particle swarm optimisation (PSO) algorithm on most datasets.

The reconstruction of shredded documents has a great significance in the fields of forensics, reconstruction of historical documents, and intelligence analysis. The reconstruction of cross-cut shredded Chinese documents is presented in this paper. The Evaluation of Match Degree is divided into two sub-problems, feature and the corresponding scoring function. A new method of the Evaluation of Match Degree which is suitable for shredded Chinese documents is presented. Information Quantity is introduced to measure the reliability of each matching, instead of regarding as the same. A novel and effective algorithm of automatic reconstruction based on Information Quantity is put forward to control the serious propagation of errors caused by the matching of shreds with low Information Quantity. Not only is the propagation of errors controlled effectively, and the error ratio reduced, but also the time complexity decreases greatly. Experiments have proven the high accuracy and superiority of the algorithm proposed in this paper.

This paper designs a scheme, which can quantize the personal privacy based on profile and context, using entropy. It aims to configure the person's privacy in numerical form that can be dynamically adjusted instead of using the traditional simple rule engine. In addition, we design the scenario to design a adjust algorithm to modify the privacy based on numerical values.

In order to improve the learning mechanism of BFNNs, the paper firstly analyzes the failure mode of BFNNs trained by SBALR, which takes the form of a local cycle. And then by mean of the sensitivity theory, a disturbance learning algorithm is developed to make the BFNNs that suffering from learning failure to escape the local cycle. The new algorithm aims to keep the existing learning performance as much as possible. Experimental results demonstrate the effectiveness of the new algorithm on both learning effect and learning efficiency.

The increasing popularity of social networks, such as online communities and collaboration networks, has given rise to interesting knowledge discovery and data mining problems. Different methods of knowledge discovery and data mining require different types of social network datasets, such as datasets including node and edge information or datasets with added labels or weights. In this paper, we review social network datasets in different formats and applications. Then, we propose a data processing method to add weights and labels for a Chinese collaboration network (C-DBLP). The weights reflect the affinity between two authors, and the labels contain information about the authors. This dataset can be used for privacy protection in social networks and social network community detection. Finally, we analyse the new data that we compiled and draw conclusions about the importance of the weights in the C-DBLP dataset in terms of social network research.

Meteorological technology has evolved rapidly in recent years to provide enormous, accurate and personalized advantages in the public service. Large volumes of observational data are generated gradually by technologies such as geographical remote sensing, meteorological radar satellite, etc. that makes data analysis in weather forecasting more precise but also poses a threat to the traditional method of data storage. In this paper, we present MHBase, (Meteorological data based on HBase (Hadoop Database), a distributed real-time query scheme for meteorological data based on HBase. The calibrated data obtained from terminal devices will be partitioned into HBase and persisted to HDFS (the Hadoop Distributed File System). We propose two algorithms (the Indexed Store and the Indexed Retrieve Algorithms) to implement a secondary index using HBase Coprocessors, which allow MHbase to provide high performance data querying on columns other than rowkey. Experimental results show that the performance of MHBase can satisfy the basic demands of meteorological business services.

In many learning tasks, there are abundant unlabeled samples but the number of labeled training samples is limited, because labeling the samples requires the efforts of human annotators and expertise. There are three major techniques for labeling the samples: semi-supervised learning, transductive learning and active learning. Semi-supervised and transductive learning deal with methods for automated exploiting unlabeled samples in addition to improve learning performance. Active learning deals with methods that assume the learner has control over the whole input space. So combing the advantage of semi-supervised learning and active learning is a practical technique for improving the learning performance. In this paper, a general framework of combing (Active Learning) AL and (Semi-Supervised Learning) SSL algorithms is proposed. Then the ensemble learning for combing AL and SSL algorithms is introduced, which is denoted by ASC (AL and SSL by Committee). At last, the ensemble learning and confidence measure of the ASC is discussed.

Privacy is one of the most important and difficult research issues in ubiquitous computing. It is qualitative rather than quantitative. Privacy preserving mainly relies on policy based rules of the system, and users cannot adjust their privacy disclosure rules dynamically based on their wishes. To make users understand and control their privacy measurement, we present a scheme to quantize the personal privacy. We aim to configure the person’s privacy based on the numerical privacy level which can be dynamically adjusted. Instead of using the traditional simple rule engine, we implement this scheme in a complex way. In addition, we design the scenario to explain the implementation of our scheme. To the best of our knowledge, we are the first to assess personal privacy numerically to achieve precision privacy computing. The privacy measurement and disclosure model will be refined in the future work.

High performance computing resources are characterized by geographically contagious distribution in chinese meteorological departments. Local meteorological bureaus are left behind by national level institutions, whether in the possession of HPC resources and application development capabilities. As a result, they fail to conduct local meteorological forecast operations or research. On the other hand, it's an urgent task for NMIC to upgrade system management from homogeneous systems to global system management across multiple heterogeneous systems. Funded by “National Science and Technology Infrastructure Plan”, regarding the requirements of resource integration, sharing and management by local and national institutions in meteorological department, a design scheme of national meteorological computing application system is proposed. The scheme utilizes computational Grid as the platform building technology. Regarding the structure of meteorological computing Grid, a distributed, tightly coupled Grid architecture is presented comprising a national level major node and subordinate nodes dispersed across the country. Within a node, the design scheme includes user interface, Grid management, high performance computer local management. Regarding integration requirement of heterogeneous computing resources, a solution is proposed consisting of globally coherent, centralized user management and meta-scheduler of job management.

Virtual Organizations (VO) are dynamic collections of individuals, institutions, and resources which play an important role in resources management. High performance computing VO scenarios are introduced to show usage of VO to manage and share resource among HPC centers. Meteorology grid computing is a new technology which aims to provide scientist with seamless, reliable, secure and inexpensive access to resources, and help scientists efficiently search and retrieve information, analyze meteorological data and predict the future weather situation. Meteorological HP computing has some typical features in accordance with the requirements for grid computing. In this paper, a layered grid architecture in Meteorology Grid Computing is proposed firstly. Regarding to the architecture, we present principles and basic concepts, architecture details and implement secondly. Thirdly, we introduce China Meteorology Application Grid (CMAGrid) System according to that mode. In this paper, integration of resources and shared service are key technologies in resources management of grid. The CMAGrid will offer meteorology knowledge environments to facilitate the forecasters and scientists with analyzing the future weather situation.

Privacy has become an important issue in Data Mining. Many methods have been brought out to solve this problem. This paper deals with the problem of association rule mining which preserve the confidentiality of each database. In order to find the association rule, each participant has to share their own data. Thus, much privacy information may be broadcasted or been illegal used. These issues can be divided into three categories: data hiding, knowledge hiding and data mining results publishing. This paper reviews the major method of privacy preserving on each category and choose some of them to complete our system. At the end, an improvement of sensitive rules hiding is proposed to make it more accuracy and security.

Privacy has become an important issue in Data Mining. Many methods have been brought out to solve this problem. This paper deals with the problem of association rule mining which preserves the confidentiality of each database. In order to find the association rule, each participant has to share their own data. Thus, much privacy information may be broadcast or been illegal used. These issues can be divided into two categories: data hiding and knowledge hiding. This paper reviews the major method of privacy preserving on each category and choose some of them to complete our system. At the end, an improvement of sensitive rules hiding is proposed to make it more accuracy and security.

Activity recognition is a hot topic in Healthcare. Machine learning is a key aspect in activity recognition. Since the number of labeled samples is limited because they require the efforts of human annotators, while the number of unlabelled data is huge because they are easy to get without human’s labeling effort. The training data is the centre of the semi-supervised based activity recognition. In this work, we emphasize the selection strategy and enlarge degree, which are the basic of the training data selection. We provide a method to utilize the available unlabeled samples to enhance the performance of activity learning with a limited number of labeled samples.