Knowledge Engineering and Management

Due to the limitation of power supply in wireless sensor nodes, the paper represented a new routing protocol which could be applied in TinyOS 2.x operating system named CTP-TICN. The new routing protocol realizes the preliminary achievement of load balance in wireless sensor network. CTP-TICN protocol introduces the “intensity of transmission” and “numbers of one hop up node” declaration, which could help nodes choose the suboptimal parent node for data forwarding on the premise of transmission stability. The simulation shows that CTP-TICN is more effective on load-balance than the CTP routing protocol and it helps the wireless sensor network to live for a longer time.

A third-order state equation with consideration of autopilot dynamics is formulated. A nonlinear coordinate transformation is used to change the state equation into the normal form. Adaptive sliding mode control theory is depicted and its stability analyses are proved. Applying this theory on the normal form equation, an adaptive nonlinear guidance law is proposed. The presented law adopts the sliding mode control approach can effectively solve the guidance problem against target maneuver and effects caused by autopilot dynamics. Simulation results show that under the circumstance of target escaping with high acceleration and big autopilot dynamics, the proposed guidance law still has high precision.

One key process in data-driven methodology is how to use the data easily and efficiently. In this paper, an immune-enhanced unfalsified controller (IEUC) is proposed to act as an efficient process to deal with data. The IEUC consists of two parts, the first part is a unfalsified controller deriving from data-driven methodology; the second part is an immune feedback controller inspired from biologic intelligent methodology. In order to examine control effectiveness of the IEUC, we apply it to a complex plant in high-speed spinning model and compare it with a simple unfalsified control scheme. Simulation results demonstrate that the IEUC can decrease system overshoot as well as reduce rising time successfully and effectively.

In this paper, we develop a real-time adaptive task allocation algorithm based on parallel dynamic coalition in WSNs. The algorithm gives a priority level to each task according to the idea of EDF. And the task with relatively higher priority will be scheduled firstly. When coalitions are parallel generated through PSO algorithm, the corresponding task of coalition will be allocated according to the current load of sensors and the remaining energy balance degree. The experimental results show that the proposed algorithm has strong capability to meet deadline constraint and it can prolong the lifetime of the whole network significantly.

As labor costs rising, development of agriculture robot to alter labor-intensive traditional agriculture is becoming more and more important .Task of transplanting is very heavy and important in bedding plants system in facility agriculture. In this paper, control strategy of transplanting healthy seedlings from high density plug tray to low density growing tray are analyzed. Cause of difficult to get the optimal solution of the whole combination route strategy in the real time, list four typical transplanting scheme, then introduced greedy algorithm to optimize control strategy. Through the simulation, greedy algorithm make the path distance shortened. Controller get approximate optimal solution at the situation of unknown optimal solution, improved the efficiency of transplanting.

This paper investigates a squeaky wheel optimization (SWO) approach to the airport gate assignment problem (AGAP). A graph coloring method is incorporated into the SWO procedure to construct solutions in our approach. Some initial experimental results are presented towards the validation of this approach.

In T-detector Maturation Algorithm with Overlap Rate, the parameter Omin is proposed to control the distance among detectors. But Omin is required to be set by experience. To solve the problem, T-detector Maturation Algorithm with NS operator is proposed. The results of experiment show that the proposed algorithm can achieve the same effect with TMA-OR when 2-dimensional synthetic data and iris data are used as the data set.

Using the continuation theorem of coincidence degree theory and analysis techniques, we establish criteria for the existence of periodic solutions of predator-prey models and impulsive perturbations. It is more appropriate to add the density-dependent term to these models in this paper. Further, computer simulation shows that our models can occur in many forms of complexities including periodic oscillation and gui chaotic strange attractor.

The principle aim of this paper is to explore the existence of periodic solution of a predator-prey model with functional response and impulsive perturbations. Sufficient and realistic conditions are obtained by using Mawhin’s continuation theorem of the coincidence degree. Further, some numerical simulations show that our model can occur in many forms of complexities including periodic oscillation and chaotic strange attractor.

Cultural Algorithms (CAs) are a series of new algorithms which depict cultural evolution as a process of dual inheritance. In this paper, cultural algorithm using Genetic Algorithms (GAs) and the knowledge in belief space to guide the evolution of population space is introduced. GAs simply use the fitness to evaluate the quality of solutions, however, it may lose the diversity of population and even lead to premature convergence. To solve this problem, we put forward a novel selection operator. Compared with conventional CA based on GA, CA with our selection operator performs better in the global convergence.

There is a close relation between the phenomenon of LOH and malignant tumor. Bicluster algorithms have been applied to the data of loss of heterozygosity analysis and can find the submatrix which is composed by SNPs loci related to cancer. But the conventional Cheng and Church method requires experience values as a threshold, and discovered results must be randomized. In this paper, we use k-means and GA to overcome this shortcoming. The experimental results demonstrate the effectiveness and accuracy of our method in discovering chromosome segments related to suppressor genes of lung cancer.

Simulated binary crossover (SBX) operator is widely used in real-coded genetic algorithms. Particle swarm optimization (PSO) is a well-studied optimization scheme. In this paper, we combine SBX together with particle swarm optimization (PSO) procedures to prevent possible premature convergence. Benchmark tests are implemented and the result turns out that such modification enhances the exploitation ability of PSO.

Clustering with background information is highly desirable in many business applications recently due to its potential to capture important semantics of the business/dataset. Must-Link and Cannot-Link constraints between a given pair of instances in the dataset are common prior knowledge incorporated in many clustering algorithms today. Cop-Kmeans incorporates these constraints in its clustering mechanism. However, due to rapidly increasing scale of data today, it is becoming overwhelmingly difficult for it to handle massive dataset. In this paper, we propose a parallel Cop-Kmeans algorithm based on MapReduce- a technique which basically distributes the clustering load over a given number of processors. Experimental results show that this approach can scale well to massive dataset while maintaining all crucial characteristics of the serial Cop-Kmeans algorithm.

Aerospace data pre-processing ground system play an important role in aerospace exploiting. Today, there’re several special aerospace data pre-processing ground system in operation. As experimental activities of space exploration continue to carry out, payload types is increasing, data types and the satellite downlink data are growing on a large scale. For multi-tasks and multi-payloads aerospace massive data processing, this article analyses and designs a aerospace data parallel pre-processing ground system with grid workflow scheduling. It can be used to create appropriate processes flexibility and quickly, adjusting the data pre-processing of multi-missions and multi-payload[2]. Addition, it has perfect scalability in business and computing ability. So, it can meet the future requirements of aerospace data pre-processing.

A universal platform is developed to provide several interfaces that every sensor can register or communicate with each other and platform (environment). To understand the operation of the platform, sensor nodes simulated by the Blackfin processor and PCs are used. This report presents the performance of the DSN based on an experiment in which different sensor nodes are connected to the platform, send/get data to/from platform and also can communication with each other. The results of this study demonstrate the principles and implementation of distributed sensor networks which can be extended to a complete and operational sensor management environment.

Knowledge-based approach for composite high-performance application building and execution is proposed as a solution to solving complex computational-intensive scientific tasks using set of existing software packages. The approach is based on semantic description of existing software, used within composite application. It allows building applications according to user quality requirements and domain-specific task description. CLAVIRE platform is described as an example of successful implementation of proposed approach’s basic principles. Exploration of described software solution performance characteristics is presented.

In distributed systems, the uncertainty of the clock drift and transmission delay is a problem that cannot be ignored, because it influences the precision of the clock synchronization directly. In this paper, considering this problem, an adaptive clock synchronization algorithm which aims the characteristics of enterprise distributed systems is proposed based on the passive algorithm. The adaptive algorithm can automatically determine the time interval of the two adjacent clock synchronization adjustments and choose the optimized measurement times and the clock adjustment value of each clock synchronization adjustment, so that, an optimized plan can be used to achieve the synchronization task. It is proved that the algorithm can inhibit effectively the influence of the uncertainty to meet the accuracy requirements of the clock synchronization in enterprise distributed systems. The algorithm has a good practical value.

Task scheduling is an essential aspect of parallel processing system. This problem assumes fully connected processors and ignores contention on the communication links. However, as arbitrary processor network (APN), communication contention has a strong influence on the execution time of a parallel application. In this paper, we propose genetic algorithms with fuzzy routing to face with link contention. In fuzzy routing algorithm, we consider speed of links and also busy time of intermediate links. To evaluate our method, we generate random DAGs with different Sparsity value based on Bernoulli distribution and compare our method with genetic algorithm and classic routing algorithm and also with BSA (bubble scheduling and allocation) method that is a well-known algorithm in this field. Experimental results show our method (GA with fuzzy routing) is able to find a scheduling with lower makespan than GA with classic routing and also BSA.

Under the third kind of constraint, using the method of OLM (Optimal Lagrange Multipliers formalism) and maximum entropy principle, we derive the statistical distribution and thermodynamic formulas in completely open system based on Tsallis entropy. Two hardships existed in the method of TMP (Tsallis-Mendes-Plastino formalism) are overcome. We give out the specific expressions of the three Lagrangian multipliers and make their physical significance clear.

In this paper, we proposed a novel symbolic controller with PID (proportional integral derivative) feedback for locally linearized system. The state-space of this system is analyzed in Brunovsky coordinates, and then a practical control structure which combines symbolic controller and PID controller is presented. Series of Experiments on an inverted pendulum are conducted, and the results show the feasibility and efficiency of our proposed structure.

In this paper, a multistage decision problem for minimizing the total cost with partially known information is considered. In each stage, a decision maker has one and only one chance to make a decision. The optimal decision in each stage is obtained based on the one-shot decision theory. That is, the decision maker chooses one of states of nature (scenario) of each alternative in every stage with considering the satisfaction of the outcome and its possibility. The selected state of nature is called the focus point. Based on the focus points, the decision maker determines the optimal alternative in each stage by dynamic programming problems.

For the under-actuated planar biped robot, we propose a novel method to change walking speed based on periodic walking. We design reference joint trajectory as Bezier curve. Based on virtual constrain, linear feedback has been used to make joint trajectory asymptotically follow the reference trajectory. To change walking speed, different Bezier parameters are chosen for different walking rate and are fused by fuzzy algorithm. As these fusion outputs are rough, PI controller is designed to regulate some legs’ parameters to get a stable walking gait. The whole control system is a two-level control structure. In the first level, a feedforward-feedback controller consisting of fuzzy controller and PI controller is used to generate reference trajectory. While in the second level, linear feedback could be used to achieve trajectory tracking. Simulation confirms the efficiency of this method.

To effectively analyze and evaluate the performances of Interbay material handling system with shortcut and blocking in semiconductor fab, an extended Markov chain model (EMCM) has been proposed, in which the system characteristics such as vehicle blockage and system’s shortcut configuration are well considered. With production data from Interbay material handling system of a 300mm semiconductor fab, the proposed EMCM is compared with simulation analytic model. The result demonstrates that the proposed EMCM is an effective modeling approach for analyzing Interlay’s performances in the early phase of system design.

Reaching ability is a kind of human sensory motor coordination. The objective of this work is to imitate the developmental progress of human infant to create a robotic system which can reach or capture objects. The work proposes to employ a double neural network architecture to implement control a robotic system to learn reaching within 3D experimental environment. A constraint releasing mechanism is applied to implement the development procedure for the robot system. In addition, the experimental results are described and discussed in this paper.

A new image recognition algorithm for a small creature—Caenorhabditis elgance is developed in this paper. This methods first use edge detection, binarization and other methods to get the body contours of C.elgance. Then using its body symmetry, C.elgance is recognized by comparing similarity of consecutive lines in image. The experimental result shows the high positioning accuracy and rapidity of the proposed algorithm. This method can also be applied to other object recognition from the biological image of creatures with symmetric body.

Approaches based on self-learning and self-organization of collective of interacting intellectual agents (sensor channels, units of neural networks structures and others DAI-forming entities) are recently more often used in development of collective recognition and control systems. Approaches to organization of co-learning and mutual learning on the basis of reliability coefficients of sensor data are described. Approach to development of recognizing multi-sensor system on a basis of two-dimensional two-level neural networks is offered. Scheme of multi-unit manipulators control on a basis of neural approach are described. The results are used for development of robotics systems for operation in extreme (underwater) conditions.

This paper is concerned with the design and analysis of a polynomial-time algorithm for an open problem in Planar Quantified Integer Programming (PQIP), whose polynomial-time solution is previously unknown. Since the other three classes of PQIP are known to be in PTIME, we also accomplish the proof that PQIP is in PTIME. Among its practical implications, this problem is a model for an important kind of scheduling. A challenge to solve this problem is that using quantifier elimination is not a valid approach. This algorithm exploits the fact that a PQIP can be horizontally partitioned into slices and that the feasibility of each slice can be checked efficiently. We present two different solutions to implement the subroutine

CheckSlice

: The first one uses IP(3), i.e., Integer Programming with at most three non-zero variables per constraint; the other is based on counting lattice points in a convex polygon. We compare the features of these two solutions.

Emotions play a critical role in increasing the believability of virtual agents. The paper develops an emotion model of virtual agent using particle system and OCC model. To better portray emotions of virtual agents, emotional experience is reflected by two aspects, which are the outer emotion defined by the intensity of particles, and the inner feeling by the number of particles. Based on the perspective of psychic energy, the interactional effect of emotions is incarnated through particle motion. Simulation is done by using Matlab software, and the results show that the emotion model can simulate better dynamic process of emotion transferring and change spontaneously.

The integral inequalities play important roles in classic measure theory. With the development of fuzzy measure theory, experts want to seek for the integral inequalities of fuzzy integral. We concern on the inequalities of Choquet integral. In this paper, Hölder type inequality and Jensen type inequality for Choquet integral are presented. As the fuzzy measure are not additive, thus what is the other conditions for integral inequalities are discussed. Besides, examples are given to show that the conditions can’t be omitted.

To solve the problems of label clustering, bad theme relevance, etc. of information retrieval especially the semantic retrieval in the context of Chinese environment. The appropriate information retrieval model for the Chinese semantic environment is constructed through introducing ontology and class label mechanism. Based on optimization of the query retrieval submitted by ontology and label to the user and through calculation of the similarity between ontology label and member engine data, the dispatching method for member search engine database is proposed; the method appropriate for extracting this semantic information retrieval model data is proposed through improvement of the traditional STC algorithm. The research on semantic retrieval technology based on ontology and label breaks through the bottleneck of semantic search and information resource management & organization and enhances the scope and quality of the semantic information retrieval.

In order to solve the problem of the knowledge base construction for a living expert system to adapt the changeable world, a set of method based on the cognitive model include the knowledge acquisition, representation, storage in organization, updating and reasoning conveniently is presented. Simulating the learning procedure of human beings is the core idea of this method from which we can find the ways how to add, delete, amend and use the knowledge in an expert system. Based on the analysis of the common procedure of children’s actions during recognizing the world, a cognitive model of concept learning is abstracted. A general concept learning algorithm, a knowledge representation method based on general rules, a logical structure in the forest shape, and a uniform data structure for storage are accordingly presented. Thus, a complete and more scientific management case for the knowledge base of expert system is provided. At last, comparing with some ontology knowledge bases, three different characteristics of this construction method are discussed.

Based on current situation that traditional manufacturing enterprises cannot combine the customer demand in supply chain when the supplier is selected, the manufacturing enterprises’ model of supplier selection based on improved QFD is proposed through analysis on the index system of supplier selection. Such model is for QFD method the important tool – “House of Quality”, which is restructured in content and form, formulates a relation matrix between customer demand and index system, and makes use of AHP method and independent collocation method for supplier selection. The final empirical analysis by examples proves that such model is feasible.

With some critical defects existing in the generation of requirement such as the method to extract concept, property and their relationship from external knowledge sources issues, we propose a novel approach to automatically generate requirement for domain ontology evolution based on machine learning theory in the light of P2P network routing model and storage characteristics of ontology resource. The method takes a comprehensive considering on term frequency, term field, similarity with original ontology and other factors to extract key concept from texts, and then through Naïve Bayes classifier compares those key concept with original ontology and extracts their relationships. We demonstrate that this research can ensure the requirement’s reliability and improve the automation and intelligent on the process of requirements generation through simulation experiments and analysis of the results.

The main problem of cooperative operation in service chain for digital library (SCDL) is resource sharing and resource unified management under heterogeneous environment. Based on the theory of Cloud Computing and its application, a service chain architecture for digital library based on Cloud Computing is constructed, and an architecture of cloud service platform for SCDL is proposed. Then a method of resource sharing based on publish/subscribe notification mechanisms and concept retrieval models is presented, and a cooperative service architecture in cloud service platform based on resource sharing is proposed. Finally, a sample of content retrieval based on Cloud is demonstrated.

The decreasing efficiency of knowledge searching and utilization caused by knowledge explosion requires the evaluation of knowledge. An evaluation algorithm based on mass knowledge evaluation was proposed and the evaluation impact brought by evaluation order, evaluation quantity and quality of knowledge was further studied through this algorithm. Grounded in this algorithm, the evaluation capability value of participants and value of knowledge was calculated by common participation action. Different individuals have different evaluation weighing of knowledge and the evaluation capability of an individual is affected by the weighing of the knowledge evaluated. The valuable knowledge can be sorted out relying the self-organization of employees which brings higher utilization efficiency of knowledge, while the knowledge evaluation capability of employees can be sorted which can be used as an index of employee performance appraisal. An enterprise knowledge management system (EKMS) was developed, which proved the feasibility of the proposed algorithm.

Nowadays decisions that affect organizations or big amounts of people are normally made by a group of experts, rather than a single decision maker. These decisions would require more than a majority rule to be well-accepted. Consensus Reaching Processes in Group Decision Making Problems attempt to reach a mutual agreement before making a decision. A key issue in these processes is the adequate choice of a consensus measure, according to the group’s needs and the context of the specific problem to solve. In this contribution, we introduce the concept of attitude towards consensus in consensus reaching processes, with the aim of integrating it in the consensus measures used throughout the consensus process by means of a novel aggregation operator based on OWA, so-called Attitude-OWA (AOWA).

Large ontology matching problem brings a new challenge to the state of ontology matching technology. In this paper we present a criteria-Based approximate matching approach to solve the problem. The main principle of our approach is based on fully explore semantic information hidden in ontology structures, either physically or logically, and tightly coupled element-level and structure-level features to obtain matching results. Through joining the quantitative structural information for the matching process can significantly decrease the execution time with pretty good quality.

The single reasoning mechanism is adopted in most of the expert system designed for fault diagnosis, the diagnosis results are single and insufficiency. In order to handle the problems which are more and more complex and improve the efficiency and accuracy, the integrated reasoning mechanism which is suitable for the route troubleshooting is designed in the paper. How to combine the route troubleshooting and the expert system by the design of integrated reasoner, the establishment of the integrated knowledge database, and the selection of the reasoning strategy and the realization of the soft ware are introduced in this paper. The route troubleshooting and expert system are combined well, the design and develop of the soft ware are completed, a diagnosis results with high accuracy, wide coverage and strong reliability can be obtained in this system, data can be conversed among three databases, then the diagnosis efficiency is improved.

Searching records in internal website is an important way for people to obtain information, however, accurate and satisfying results cannot be acquired easily with conventional methods. In order to improve the effectiveness of search algorithm, in this paper we adopt ontology technology to the searching process and build up an ontology model for searching. Finally, some corresponding tests were carried out in order to valid the performance of the algorithm.

Feature modeling technology makes micro device design in a more intuitive and efficient way. To support the feature-based information organization and rule-based features mapping procedure, a design framework and key enabling techniques are presented. Firstly, on the basis of hierarchical feature structure, the architecture for the model construction and optimization is constructed. Secondly, in accordance with the ontological representation, the knowledge of different features is hierarchically organized. Then, the feature mapping relationship is built with the rule-based reasoning process. By ontology-based features representing framework and reasoning procedure, the constraint features and associate rules are connected with other features. Finally, the design model and manufacturing model are optimized simultaneously.

This paper studies a supply chain network design problem with stochastic parameters. A Value-at-Risk (VaR) based stochastic supply chain network design (VaR-SSCND) problem is built, in which both the transportation costs and customer demand are assumed to be random variables. The objective of the problem is to minimize the allowable invested capital. For general discrete distributions, the proposed problem is equivalent to a deterministic mixed-integer programming problem. So, we can employ conventional optimization algorithms such as branch-and-bound method to solve the deterministic programming problem. Finally, one numerical example is presented to demonstrate the validity of the proposed model and the effectiveness of the solution method.

This paper puts forward a task-driven knowledge reuse framework for complex product design process. Domain ontology is used in the framework, which is fundamental for customization of the systems developed under this framework. The task representation model, as part of the domain ontology, is discussed in this paper; then a mechanism of sharing knowledge among instances of the same task class or between tasks which have some special relationships is discussed. The algorithm for generating knowledge learning schedule is also given based on the task relationship and the dependency of knowledge items. Through this framework, the task instances containing historical experiences of input and output can be stored and retrieved seamlessly during the execution of tasks without manual efforts.

Lyrics can be used to predict the emotions of songs, and if combined with methods based on audio, better predictions can be achieved. In this paper, we present a new approach to lyric emotion regression. We first build a Latent Dirichlet Allocation (LDA) model from a large corpus of unlabeled lyrics. Based on the model, we can infer the latent topic probabilities of lyrics. Based on the latent topic probabilities of labeled lyrics, we devise a scheme for training and integrating emotion regression models, in which separate models are trained for latent topics and the outputs of those models are combined to get the final regression result. Experimental results show that this scheme can effectively improve the emotion regression accuracy.

This paper presents an online traffic system to recommend taxi drivers the fastest-path of picking passengers up. Several systems have been studied to find and recommend the shortest-paths on distance in mobile scenarios. However, in practical traffics, we discover that the shortest-path is usually not the fastest-path due to congestion. Especially for the taxi drivers, the fastest-path to pick up passengers is the best choice. Analyzing a real trace data including about 2000 taxis in a 22 square kilometers area in 7 days in Shanghai. Then we design a practical recommendation system to process the fastest-path selection. Experimental results show that our online system can quickly recommend the almost exact fastest-paths to taxi drivers for picking up passengers in real traces.

This essay firstly introduces a survey of computer dynamic forensics, and then proposes a computer dynamic forensics system model. Aiming at solving the problems that the computer dynamic forensics faces in the data analysis stage, we apply the Outlier Detection to maguuiimous data analysis in the computer dynamic forensics. We apply this technique on KDDCUP99 data set and get satisfactory results.

We present a novel unsupervised method for sentence compression which relies on a Stanford Typed Dependencies to extract information items, then generates compressed sentences via Natural Language Generation(NLG) engine. An automatic evaluation shows that our method obtains better results. We demonstrate that the choice of the parser affects the performance of the system.

In this paper, we analyzed the shortcoming of Feature Weighted SVM and Sample Weighted SVM, then a new SVM approach is proposed based on the comprehensive feature and sample weighted. This method estimates the relative importance (weight) of each feature by discernibility matrix. It utilizes the weights for computing the inner product in kernel functions. In this way the computing of kernel function can avoid being dominated by trivial relevant or irrelevant features. Then we estimate the weight of each training samples by the feature weight and similarity between samples, in order to reduce the influence of non-critical samples and noise data on the SVM learning and improve the noise immunity. Experimental results show that comparing with simply considering the importance of feature or sample, the proposed method can more effectively improve the classification accuracy of SVM.

It is a key technology in online examination system that automatic correcting. Semantic similarity is the main way to solve the auto-correcting, but the exact calculation depends on the words similarity of the student’s response and accurate answers. They are the important reasons to decrease the accuracy of word segmentation that identify the new words (Glossary) and segmentation ambiguity problem. A new method to construct dictionary having glossary is proposed that the new words are identified by PAT array and ambiguities are eliminated by association rule mining. The accuracy of segmentation may achieve 95% and have 5% of increase in automatic correcting of computer organization.

This paper established the structural database using the method of Threat and Error Management (TEM), and obtained the characteristics of three types of data which are classified by threat, error and unexpected situation. Based on TEM frame, Neural Network is applied to add data, and Bayesian Network is applied to study the correlation among threat, error and unexpected situation. Here comes the conclusion: 1) Through applying Bayesian Evaluation to 625 selected samples, we found that specific unexpected situations have high correlation with some specific threat and error, Bayesian Evaluation reveals that when some unexpected situation happen, the high correlated threat and error to the unexpected situation. For instance, the threat and error that have highest correlation to in air unexpected situation are: Procedure Internal threat and communication error among Air Traffic Control and aircrew. 2) The high occurrence of some threat and error don’t necessarily have high correlation with some high unexpected situation; high probability of some threat and error don’t necessarily lead to unexpected situation. Research achievements provide data analysis skill to Air Traffic Control on safety management control, who could make corresponding prevention measure.

Accuracy of part-of-speech tagging is critical to downstream sub-tasks in front-end text analysis model of text-to-speech System. Uyghuris an agglutinative language in which numbers of words are formed by suffixes attaching to a stem (or root). Owing to there are unlimited new formed and derived syntactic words in Uyghur, Sizes of part-of-speech tagging set were big and out-of-vocabulary words often occurred in conventional Uyghur part-of-speech tagging method which directly trained and predicted the part-of-speech of word. To address this problem, this paper proposes the idea that trains the part-of-speech of stem and predicts the part-of-speech of word mainly by stem. Bi-gram language model is used to segment the stem and affix boundary of word, hidden markov model is used to train and predict part-of-speech of stem. In the end, rule adjusting method is used to adjust the changed part-of-speech of word when suffix attaching to a stem. Experimental result shows that proposed method obviously reduces the part-of-speech tagging error rate comparing to conventional part-of-speech tagging method.

This paper proposes a method to identify scenes in texts with relevant measures, namely to infer the locations where actions take place without any explicit description in texts. Firstly, 48 scene categories are classified by the sememe hypernym and concept similarity on HowNet. Secondly, the hierarchical scene information tuples generated on action tuples and scenes from corpus are extracted by calculating the relevancy for each tuple, and then the knowledge base is generated. Finally, the constructed knowledge base is used to identify scenes in novel texts, and the experimental results on 1-best and voting shows the effective of the method.

A rule-based approach for Chinese zero anaphor detection is proposed. Given a parse tree, the smallest IP sub-tree covering the current predicate is captured. Based on this IP sub-tree, some rules are proposed for detecting whether a Chinese zero anaphor exists. This paper also systematically evaluates the rule-based method on OntoNotes corpus. Using golden parse tree, our method achieves 82.45 in F-measure. And the F-measure is 63.84 using automatic parser. The experiment results show that our method is very effective on Chinese zero anaphor detection.

Blogs are user generated content discusses on various topics. For the past 10 years, the social web content is growing in a fast pace and research projects are finding ways to channelize these information using text classification techniques. Existing classification technique follows only boolean (or crisp) logic. This paper extends our previous work with a framework where fuzzy clustering is optimized with fuzzy similarity to perform blog classification. The knowledge base-Wikipedia, a widely accepted by the research community was used for our feature selection and classification. Our experimental result proves that proposed framework significantly improves the precision and recall in classifying blogs.

Materialized views are the most common approach that can provide optimal performance in processing time, especially for

OLAP

queries known for their great complexity. Due to the large computation and storage limitation, materialization of all possible views is not possible. Therefore, the key issue is to choose an optimal set of views to materialize. However, this task is a very hard, especially in the data warehouses context, where a trade-of-between performance and view storage cost must be taken into account when deciding which views should be materialized. Addressing this problem, we propose a new approach with two main phases. The first involves pruning the search space to reduce the number of views candidates. In this order, we use a distributed clustering approach using multi agents system that can significantly reduces the complexity of the selection process The second phase uses also a multi agent’s architecture to capture the relationships between views candidates to select the final set of materialized views. This set minimizes the query processing cost and satisfy the storage constraint. We validate our proposed approach using an experimental evaluation.

On-line news event detection is detecting the first news report of a news event from various news sources in real time. Related to on-line news event detection, in this article, the author firstly introduces a news representation method for the news factors modeling based on the time, locations, characters (or organization), contents, and so on, and deducing a method related to the features of different types of news factors to calculate the weight of those news factors. Considering the insufficient of the traditional detection algorithms, then the author presents the algorithm of Micro-clusters-based on-line news event detection with Window-Adding and conducts an experiment based on news data which is collected in reality. The author achieved a satisfied experimental result verifying the validity of the proposed method.

This paper presents a novel method to recognize Chinese unknown word from short texts corpus, which is based our observation of both single-character and affix models of Chinese unknown word. In our approach, we collect some news titles of a news site and view these titles as short texts. There are three steps in our approach: First, all collected news titles are segmented with ICTCLAS, and statistics of potential unknown words are conducted. Second, all potential unknown words are classified into either single-character model or affix model based on structures of unknown word. Some filtration methods are used to filter garbage strings. Finally, unknown word is extracted according to the frequencies of word. We have got the excellent precision and the recalling rates, especially for the single-character model. The experiment results show that our approach is simple yet effective.

The popularity of microblog brings new characters to information diffusion in social networks. Facing new challenges of understanding information propagation in microblog, the framework of information producing and receiving was proposed. A general model named competing-window is also presented based on human behavior. The detailed composition of the model and its basal mathematical description are also given. In addition, a parameter called information lost as a supplement to measure dynamics of information diffusion. Meanwhile, the further application of our model to information processing and propagating was pointed out. All those work is based on the studies on human dynamics. Finally, to verify applicability, the model was applied to empirical data crawled from Sina-weibo. The interesting patterns extracted from empirical data indicate that microblog in deed is fundamentally characterized by human dynamics.

An inconsistency-based feature reduction method is firstly proposed in this paper, based on which a simple and direct rule extraction method for data classification is addressed. Because the proposed classification method depends directly on data inconsistency without leaving any value contents of datasets, it could be utilized to build a novel privacy-preserving scheme for isomorphic distributed data. In this paper, a simple privacy-preserving classification model is proposed based on the inconsistency-based feature reduction and its direct rule extraction method. Experimental results on benchmark datasets from UCI show that our method is both in good correctness performance and model efficiency.

When mobile users retrieve interested location information through location based service (LBS), they should always provide their accurate location, which may leak their location privacy. In order to prevent privacy leakage when service provider is not reliable, some solutions based on two ties architecture are proposed. However, these solutions are either hard to implement or may lead to huge communication cost. In this paper, we present a novel technique, called Guess-Answer. Through the interaction between user and server, the location information within a region that satisfies the minimum privacy requirement of the user would be delivered. Through proper analysis and experimental study, we demonstrate that Guess-Answer is an efficient and effective way to achieve personalized privacy.

Semantic-based document clustering has been a challenging problem over the past few years and its execution depends on modeling the underlying content and its similarity metrics. Existing metrics evaluate pair wise

text similarity

based on text content, which is referred as

content similarity

. The performances of these measures are based on co-occurrences, and ignore the semantics among words. Although, several research works have been carried out to solve this problem, we propose a novel similarity measure by exploiting external knowledge base-Wikipedia to enhance document clustering task. Wikipedia articles and the main categories were used to predict and affiliate them to their semantic concepts. In this measure, we incorporate context similarity by constructing a vector with each dimension representing contents similarity between a document and other documents in the collection. Experimental result conducted on TREC blog dataset confirms that the use of context similarity measure, can improve the precision of document clustering significantly.

Blogging has gained popularity in recent years. Blog, a user generated content is a rich source of information and many research are conducted in finding ways to classify blogs. In this paper, we present the solution for automatic blog classification through our new framework using Wikipedia’s category system. Our framework consists of two stages: The first stage is to find the meaningful terms from blogposts to a unique concept as well as disambiguate the terms belonging to more than one concept. The second stage is to determine the categories to which these found concepts appertain. Our

Wikipedia based blog classification

framework categorizes blog into topic based content for blog directories to perform future browsing and retrieval. Experimental results confirm that proposed framework categorizes blogposts effectively and efficiently.

Traditional automatic machine translation (MT) evaluation methods adopt the idea that calculates the similarity between machine translation output and human reference translations. However, in terms of the needs of many users, it is a key research issues to propose an evaluation method without references. As described in this paper, we propose a novel automatic MT evaluation method without human reference translations. Firstly, calculate average

n-grams

probability of source sentence with source language models, and similarly, calculate average

n-grams

probability of machine-translated sentence with target language models, finally, use the relative error of two average

n-grams

probabilities to mark machine-translated sentence. The experimental results show that our method can achieve high correlations with a few automatic MT evaluation metrics. The main contribution of this paper is that users can get MT evaluation reliability in the absence of reference translations, which greatly improving the utility of MT evaluation metrics.

The increasing volume of bulk unsolicited emails (also known as spam) brings huge damage to email service providers and inconvenience to individual users. Among the approaches to stop spam, Naive Bayes filter is very popular. In this paper, we propose the standard Naive Bayes combining with a

tri-gram

language model, namely TGNB model to filter spam emails. The TGNB model solves the problem of strong independence assumption of standard Naive Bayes model. Our experiment results on three public datasets indicate that the TGNB model can achieve higher

spam recall

and lower

false positive

, and even achieve better performance than support vector machine method which is state-of-the-art on all the three datasets.

Sponsored Search has evolved as the delivery of relevant, targeted text advertisements for Web queries. To match the most relevant advertisements for queries, query expansion algorithms were deeply researched during previous works. While most of current state-of-the-art algorithms appeal to Web search results as external resources to expand queries, we propose a novel approach based on Wikipedia for query augmentation against rare queries in sponsored search. By retrieving the top-

k

relevant articles in Wikipedia with Web query, we can extract more representative information and form a new ad query for the web query. With the new ad query, more relevant advertisements can be identified. To verify the effectiveness of our

wiki-based

query expansion methodology, we design a set of experiments and the results turn out that our approach is very effective for rare queries in sponsored search.

With the development of the service technology, more and more organizations are publishing their business function as service through Internet. Service classification is a key approach for service management. With the quick increase of service number, the cost of classifying these services through manual work is becoming more and more expensive. A service automatic classification approach based on WordNet by combining text mining, semantic technology and machine learning technology was given in the paper. The method only relies on text description of services so that it can classify different type services, such as WSDL Web Service, RESTfulWeb Service and traditional network based software component service. Though text mining and applying word sense disambiguation models, a service can be described as a sense vector with no ambiguous. Then a K-means algorithm is used to classify these services. Experimental evaluations show that our classification method has good precision and recall.

Focusing on the problem fuzzy analytic hierarchy process (FAHP), this paper investigates the inconsistency problems of preference information about alternatives expressed as a fuzzy complementary judgement matrix by a decision maker. Firstly, the average inconsistency index is selected to check whether the matrix meets the requirement of the consistency. Then, transform equation with a variable parameter is proposed after a fully consider of compatibility, and, a linear programming model is constructed under the constrain of both satisfaction consistency and compatibility to figure out the most suitable value of the variable parameter. Finally, the original matrix is transformed to an acceptable one, and the new method is illustrated with two numerical examples.

An approach has been proposed to enhance the performance of tightly coupled GPS/SINS integrated system, with applying phase-adjusted pseudorange algorithm (PAPA), which is dedicated to decrease pseudorange noise with the aid of high precise carrier phase optimally, based on the extended Kalman filter (EKF). The theory of optimal PAPA is described first, followed by the construction of tightly coupled GPS/SINS integration system as well as the Kalman filter. Relevant experiments including both static and dynamic ones have been accomplished to verify the validity of the approach proposed. The experiment results demonstrated prominent promotion with the EKF-based PAPA, compared with conventional tightly coupled integrated algorithm.

In order to improve the efficiency of spatial data access and retrieval performance, an index structure is designed, it solves the problem of low query efficiency of the single index structure when there are large amount of data. Through the establishment of correspondence between the logical records and physical records of the spatial data, the hybrid spatial data index structure is designed based on 2

K

–tree and R-tree. The insertion, deletion and query algorithm are implemented based on the hybrid tree, and the accuracy and efficiency are verified. The experimental results show that the hybrid tree needs more storage space then R-tree, but with the data volume increasing the storage space needed declining relatively, and the hybrid tree is better than the R-tree in the retrieval efficiency, and with the data volume increasing the advantage is more obvious.

Building domain ontologies is an important topic in semantic Web researches. The GuTao method that is a kind of the formal concept analysis is used in this paper for the problem. Because the GuTao method can only process single-value attributes, the interest rate is introduced as a basis of the conversion from multi-value attributes to single-value ones. An improved GuTao method is proposed to build domain ontologies. In this method domain ontologies is represented as derivative concept lattices in a formal background to make them complete and unique. Its building process is based on rigorous mathematical foundation.

In order to solve the problem of hardly keeping good performance on data presentation and usability of the data warehouse, a hybrid-driven data warehouse modeling method is proposed based on ontology. Firstly, a data-driven method is established using ontology to define the potential facts and dimensions. Furthermore, a demand-driven method is proposed based on the analysis of business process and specific business requirement to clear the fact and dimensions. Finally, with the analysis of medical domain and the specific business requirement, a data warehouse modeling about medical domain and data mining are used as a case study to verify the correctness and feasibility of the method. The result shows that the method can meet the medical domain requirements of data mining and provides a valuable reference to presenting data.

To strengthen the research on Traditional Chinese Medicine theory and the knowledge representation of syndrome. Extension set is chosed to symbolize the syndromes of Gastric disease and Gastric Precancerous Lesion. After Extenics conjugate analysis of Model of Gastric Precancerous Lesion, overall evaluation elements is established and framework of risk assessment of gastric cancer in traditional Chinese medicine syndrome formed. Finally we establish Classical domain and section domain so as to calculate the exact risk for Gastric Precancerous Lesion changing into Gastric Cancer.

This paper discusses relational algebra extended by recursive relations (tables). The interpretation of the recursive table is proposed, the closure of the extended relational algebra is proved, and a new approach to modeling a physical database structure is offered that is suitable for representing complicated hierarchical data sets. It combines methods of set theory for the recursive relations within the framework of a single modeling paradigm, which allows users to define self-similar, partially self-similar, or hierarchical sets. The use of recursive relations in the definitions of self-similar objects yields representations that can be rendered at varying levels of detail or precision at run time.

For distributed relational data sources in oilfield domain, we propose a semantic integration method of building virtual knowledge base using ontology mapping technology. In this method, two kinds of mappings are created and two ontologies designed specially to record these mappings, by which semantic query can be rewritten into SQL statements to be executed on data sources, and results will be returned as knowledge instances. Moreover, a knowledge exchange center is provided to cache some results to improve query efficiency. Since query results directly come from data sources rather than by instances transformation, this query-driven integration system can provide a unified, virtual knowledge base which can be served to achieve decision support for oilfield production.

The paper combined seven-step method and the skeleton method of ontology constructions, and proposed computer domain ontology construction method. On the basis of computer domain ontology, the paper discussed semantic retrieving technology and finally, we design and implement a computer literature retrieval system prototype which has certain reasoning and semantic expansion capabilities.

In this paper, a new algorithm of dynamic display is introduced for design of the LED screen; it put forward drum type data organization way for any size display area LED screen. Illustrated the Corresponding relationship between LED display area X, Y coordinators and memory storage unit i, j byte address. Dynamic LED screen rolling display arithmetic and programming are completed. In order to verify the effectiveness of our new algorithm, a LED display system is designed and implemented with the single chip microcomputer VRS51L3074. The result shows that this algorithm improves the performance of dynamic display.

In this paper, we shall first briefly introduce Knowledge Management System (KMS) and a background on the Recognizing Textual Entailment (RTE) that is the theoretical basis for our redundant and conflict detection (RCD). Based on the above concepts, a framework of redundant and conflict detection in KMS will be introduced, and some common information retrieval and RTE methods will be used in our framework. Finally, we implement our framework on an existing KMS and have an acceptable result.

According to the idea of Kauffman’s NK model, and considering the meaning of the measurement of complexity that a system possessing, a method of how to measure the complexity of a system was built. Here, a system was divided to three parts: itself, input and output. Each part had its elements, and among all the elements of the three parts there are relations, so the matrixes can be built based on the relations, therefore, with the idea of NK model, the complexity can be calculated through the matrixes.

On the uplink MC-CDMA channel estimation of multi-carrier, the thesis proposes a new channel estimation scheme with the aided of the estimated channel parameters. Compared to conventional uplink MC-CDMA channel estimation, the proposed scheme can estimate channel paratemeters accurately, including estimation path numbers and time delays. The proposed scheme successfully resolves the robust and reduce the computational complexity when the conventional scheme determines the important taps by threshold. The simulation result proved the proposed scheme.

A kind of linguistic truth-valued intuitionistic fuzzy lattice algebra based on the point view of intuitionistic fuzzy set and linguistic truth-valued lattice implication algebra is discussed. As a fundament of linguistic truth-valued intuitionistic fuzzy, some algebra properties are obtained. The result shows that linguistic truth-valued intuitionistic fuzzy lattice algebra is a residual lattice, but it is not MTL-algebra and

R

0

-algebra.

Formal engineering constitutes a very important issue in software engineering projects in real life. The developing of software does not always reach the desired level of reliability and performance even the life cycle of the project used to be controlled by methodologies and specific tools as Formal Languages and Formal Methods. Despite the efforts the question is that even in the best cases, the final product has a lot of errors and in some cases these errors produce catastrophic disasters (ARIANE 5, for example). This paper shows a new proposal to formalize the life cycle according to the worker at each stage and the importance of using not only the classical logics (propositional and first order), but also fuzzy logic in order to formalize the certain and uncertain information involved in natural language.

A computer system is a network of devices working together. Devices are in constant change during the life cycle of the system. Replacing devices is one of the main aims of systems performance optimization. Assuming a system contains several uncertain variables and taking into account some system evaluation indexes such as loss function, risk, uncertainty distribution, etc., this article deals with the problem of measuring performance, risk and reliability of computer systems simultaneously. A tool for the evaluation of these three factors has been developed, named EMSI (Evaluation and Modelling of Systems). It is based on an uncertainty multicriteria decision-making algorithm. Several implemented functionalities allow making comparisons between different devices, evaluating the decision of including or not a new unit into the system, measuring the reliability/risk and performance of an isolated device and of the whole system. This tool is already working at Complutense University of Madrid.

In this paper, the empirical estimates is derived for the parameter of the exponential model based on record values by taking quasi-prior and inverse prior Gamma distributions using the LINEX loss and entropy loss functions. These estimates are compared with the corresponding maximum likelihood (ML) and Bayes estimates, and also compared with the corresponding Bayes estimates under squared error loss function. A Monte Carlo simulation is used to investigate the accuracy of estimation.

This paper presents new methods that have been developed for registration of gray scale image, these methods are based on improve the registration precision and the ability of anti-noise, the image registration process was based on analyzed the situation of the point set of histogram, and defined the formula of the histogram divergence. To speed up searching the registration parameters, all were done in the wavelet field and a hybrid algorithm based on genetic algorithm and Powell’s method was used to optimize this parameters. Experimental results proved the algorithm can apply wider optimization methods and have better anti-noise robustness performance.

The preservations of location independent riskier order under certain monotone transformations are investigated. It is also showed that location independent riskier order can preserved formation of order statistics.

QuEF is a framework to analyze and evaluate the quality of approaches based on Model-Driven Web Engineering (MDWE). In this framework, the evaluation of an approach is calculated in terms of a set of information needs and a set of quality characteristics. The information needs are requirements demanded by users of approaches. On the other hand, the quality characteristics are specific aspects that the approaches provide to their users. In these lines, there is a gap in the importance of each quality characteristic in the QuEF and the degree of coverage of each information need regarding the quality characteristics. In this contribution, we propose a method to define the Quality Model within QuEF. This method is based on the Analytic Hierarchy Process in order to establish the importance of the quality characteristics and the degree of coverage of each requirement of the information needs regarding the set of quality characteristics. Furthermore, a software application that develops the proposed method is presented.

This paper aims at specifying procedures and patterns for developing high-quality reusable requirements and for engineering new requirements documents with reuse. Formalization of the requirements is a promising approach to validating requirements in the reuse process. It improves the consistency, correctness, and completeness of reusable requirements. This paper reviews the current research on reuse and requirements formalization. Based on a strategy that introduces formalization into systematic requirements reuse, we present patterns for activities on formalization-based requirements reuse.

In our recent paper “Criteria-Based Approximate Matching of Large-Scale Ontologies (Liang et al., 2011)” we had the following corrections. We adjusted part of the article without changing the basic content of the paper. We deleted the improperly cited formulas in Section 3 and added some explanatory text in Section 4.

No.

Change

Addition

1

Deleted from P284 line 29 “The formula of node global density Gden (c) is:” to P285 line “distance effects. dis(c, N) is the length of shortest path from c to N.”

Replaced with “The node global density is depend on the number of node c’s direct subclasses, direct superclass and functional relations. Different types of link play different role in density calculation. The node local density criterion favours the densest concept in a local area, for being potentially the most important for this particular part of the ontology.”

2

Deleted from P285 line 14 “formula (4)” to line 17 “formula (5)”

Correspondingly modified the formula label referenced in the text.

3

Added “[4]” in the end of P285 line 19

4

Deleted from P285 line 20 “We define Coverage( S) as the measure of the level of coverage of a set of concepts” to line 32 “formula (8)”

5

Modified P286 “formula (9)” to “formula (2)”, “formula (10)” to “formula (3)”, “formula (11)” to “formula (4)”

Correspondingly modified the formula labels referenced in the text.

6

After P286 line 18 “4 Ontology Modular Partitioning”, added a new paragraph text: “The objective of ontology partitioning is to partitioning monolithic large ontology into a set of significant and mostly self-contained modules in order to allow its easier maintenance and use. We propose a method for automatically partitioning the set of ontology vertices into a set of disjoint clusters. The structural proximities among the vertices in a cluster are high; while those coupling crossing different clusters are low. Each cluster is a sub-part of ontology and the union of all clusters is semantically equivalent to the original ontology O. The partitioning algorithm proposed in this paper is modelling the ontology corresponding hierarchical concept network to complex electric circuit.”

7

Modified P286 “formula (12)” to “formula (5)”, “formula (13)” to “formula (6)”

Correspondingly modified the formula labels referenced in the text.

8

Modified P287 “formula (14)” to “formula (7)”, “formula (15)” to “formula (8)”, “formula (16)” to “formula (9)”, “formula (17)” to “formula (10)”

Correspondingly modified the formula labels referenced in the text.

9

After P287 line 7, added a new paragraph text: “In general Eq. (9) takes O(n3) time to solve a set of equations. However, we can actually cut the time down to O(V +E). We first set V1 = 1; V2 = …= Vn = 0 in O(V) time. Starting from node 3, we consecutively update a node’s voltage to the average voltage of its neighbours, according to Eq. (6). The updating process ends when we get to the last node n. We call this a round. Because any node i has ki neighbours, one has to spend an amount of O(ki) time calculating its neighbour average, thus the total time spent in one round is

${\rm O}(\sum\limits^{n}_{i=3}k_i)$

= O(E). After repeating the updating process for a finite number of rounds, one reaches an approximate solution within a certain precision, which does not depend on the graph size n but only depends on the number of iteration rounds. So no matter how large the graph is, so the total running time is always O(V + E).”