Abstract:Mining association rules and sequential rules from large databases is an important task of data mining. Previous work is focused on definite and accurate concepts, which may not be concise and meaningful enough for human experts to easily obtain nontrivial knowledge from the rules discovered. The definition of fuzzy concepts is based on fuzzy set theory, which is especially useful when the discovered rules are presented to human experts for examination. In this paper, algorithms are presented for discovered rules are presented to guman experts for examination.In this paper,algorithms are presentrd for discovering fuzzy association rules and fuzzy sequential rules expressed by fuzzy concepts from large relational databases.

Abstract:A procedural model is presented in this paper for generating water in the stereo animation. Here the emphasis is not on a realistic water animation, but a stylistic representation of the water as used in cartoon. In this modeling approach, hand drawn water series are decomposed into stationary and moving components to abstract static and dynamic structures associated with water forms. Based on those structures, a hierarchical framework is constructed to generate 2D animation series that match those hand-drawn series.The model offers flexible control overthe structure,shape,color,timing etc in a sequence of water effects to produce some other canrtoon such as water jet by additional level in the hierarchical framework.A significant contri-bution of this work is the extension of cartoon water effects to 3D,thus stereo cartoon water animations can be generated with wase in the CAVE by use of the bepth information in the 3D cartoon water model.

Abstract:In this paper, the authors mainly intend to clarify the relation between NP and PP . A randomized version of NP is given. Based on this equivalent definition of NP , another randomized complexity class is given: SUPER-NP . Although the SUPER-NP is very close to NP , but it is found surprisingly that PP?SUPER NP and thus NP?PP?SUPER-NP . In light of NP=PCP(log, O(1)) and the closeness of NP and SUPER-NP it is hoped that PP?PCP(log²,O(1)) conjecture can be peoved by showing that SUPER-NP?PCP(log²,O(1)).

Abstract:This paper describes an Ada based object oriented modeling language AML which takes a unique and innovative approach to import the fundament and philosophies of Ada95, and extends it with some facilities for the support of object oriented modeling so that AML is suitable to model large software systems. AML adopts the package concept from Ada95 and makes it become the core construct of AML. At the same time, AML introduces the facilities such as the task unit and the protect unit to describe concisely positive control com-ponents and resource protect compoents.Absorbing the idea of multiple viewpoint models,AML disjoints the information describing different characteristics of the given entity.AML uses the new concurrency model and the restriction facility to address modling concurrencyand nonfunctional characteristics of systems.Also AML has good extensibility and can be applied to all kined to all kinds of domains.In short,AML is user-oriented,developer-oriented,and system-oriented modeling,and overcomes some limitations such as insufficient expressive capability and limited application domain of some other modeling languages.

Abstract:Most genetic algorithms do not use the knowledge in the related problem fields completely when searching the approximate solutions. A new kind of genetic algorithm with modified fitness functions the presented in this paper. In this algorithms, both the function value at the searching point and the function change rate at the point are combined into fitness functions. It makes the chromosome code chosen by probability be able to have both smaller function value (for minimum problem) and higher function change rate. The experimental results show that the new algorithm is convergent much faster than the standard genetic algorithm is.

Abstract:Parsing the logical embedded complex sentence in Chinese consists of recognition of both logical lelation and logical level. Recognizing of logical level, which is to parse the level in this kind of sentences with lots of logical relations, is the aim in this paper. To formalize effectively the logical level in those complex sentences, in this paper, the concept of relation hierarchical tree is defined, and in view of this concept, the grammar is designed to construct the parser for this kind of complex sentences with shift-reduce algorithm improved. The parser in this paper is evaluated in open test. The accuracy, 93.56%, shows both the effectiveness of the theory in this paper.

Abstract:CBSD (component based software development) becomes one of the mainstreams of software development paradigms. The essential aspects it concerns are standardization of component models and interoperability of components. Recently it is widely accepted that interactions, as non functional properties, should be separated from functional components. As a specific part governing the coordination of components, interaction protocols become one of the focuses in CBSD paradigm. Based on OSD (object sequence diagram) notations in UML (unified modeling language), an approach for specification and analysis of component interaction protocols is achieved in this paper, In this approach, not only event causality based formalism is defined for specification of interaction protocols, but also some syntactic and dynamic analysis techniques, such as model checking, are provided for verification of OSD specification. Moreover, a framework for development of interaction protacols is sketched out, in which visual modeling and formal analysis features the approach especially.

Abstract:For developing mature software frameworks and components, it is necessary to elicit and incorporate software patterns in the process of software development. Based on software patterns, the object\|oriented software development method is discussed in this paper. Based on the analysis of requirement specifications for business application domains, an analysis approach and a basic role model of software framework are presented in this paper. An analysis pattern of framework architecture, basic role classes and their structure are elicited, an abstract factory patten for data access is designed. This method can apply to developments of application software frameworks and components.

Abstract:This is an overview of protocol research in electronic commerce. Electronic commerce protocols are the technical foundation to carry out the electronic commerce. The principles for designing electronic commerce protocols such as security, anonymity, atomicity, nonrepudiation and transaction size are introduced. Some famous existing electronic commerce protocols are described and analyzed as well.

Abstract:The features of skin color are wide used in the areas of MPEG4, content based image retrieve and virtual reality. Presently, there are plenty of the methods of the skin color clustering and of the varieties of the clustering results. Some elementary problems of the features of skin color, such as the effect of hight source, of color constancy and of color space in which the skin color clustering is performed, are investigated in this paper. A new color space is constructed for the skin color clustering and correlation between the mage data and the color of iluminants is introduced. It is proved that high lights and shading only changes the saturation of the skin color. In this framework, both the results of the skin clustering and their applications in the facial detection and hand gesture recognition in complex background are presented.

Abstract:In real-time applications, such as IP phones and remote conference systems utilizing RTP (real\|time transport protocol) for transmitting continuous media data over Internet, packet loss that impairs quality of transmitting severely is inevitable. It is a better way to solve this problem by adding a function of lost repairing to RTP. In this paper, a method of lost repairing with FEC (forward error correction) based on RTP and an RTP library, which was designed and implemented based on this method are introduced. Then a testing method of lost repairing function and the results are illustrated.

Abstract:In this paper, some methods are proposed to discriminate utterances from the speech signal. A corpus containing emotional speech of happiness, anger, surprise and sorrow with over 300 utterances from five speakers is recorded. Ten emotional features are extracted from the speech signal. Three emotion recognition methods are introduced based on principal component analysis. Using these methods, recognition performance is obtained, which is close to human performance on the task.

Abstract:Collision detection is of significant meaning in many fields such as computer graphics, CAD, simulation and virtual reality. Bounding volume hierarchy provides an effective method to resolve the intrinsic time complexity in collision detection. In this paper, a method is proposed based on FDH (fixed direction hul) for collision detection and illustrate that FDH is applicable to exact collision detection in complex environments, as well as in deformable environments. The comparison of FDH and other bounding volumes are also prsented through experimental data.

Abstract:In the multi-processor environment, real-time systems have a concurrent event-driven nature. The software structure presents multiple synchronization points and producer-consumer interrelations, which give rise to a very complex control structure. The lack of standard methods and tools for software development in these systems results in poorly specified software with an unclear structure that is very difficult and expensive to maintain. In order to develop this kind of software with high efficiency and maintainability, the Petri net is adopted to represent multi-processor system software since the Petri net is a very powerful description tool for the parallel and concurrent program. In this paper, a methodology for programming multiprocessor real-time systems is discussed. This methodology is based on two programming levels: the task level, which involves programming the basic actions that may be executed in the system as units with a single control thread, and the job level, on which parallel programs to be executed by the complete multiprocessor system are developed. The model that has been chosen for the representation of the system software is based on an objected oriented oriented Petri net, which facilitates job-level Programming. In this research, in order to support a reliable program structure, a natural, simple, and flexible procedure has been sought to describe the concurrence, synchronization, and communication phenomena inherent in this kind of software in a way that is very close to how the human programmer conceives of them. The structure and implementation of an operating system for the proposed methodology are described in this paper.

Abstract:The design and implementation of a personalized information retrieval agent, DOLTRI-Agent (distance and open learning training resource information agent), is introduced in this paper. DOLTRI-Agent is to retrieve training resource information for DOL (distance and open learning). It is based on the multi agent architecture, including three sub agents and three other components. DOLTRI-Agent system owns the autonomous functions of information retrieval, maintenance and updating. Furthermore, it has the feature of personalization: it can continuously learn user knowledge, such as user interested information fields and user's babits; can also utilize this knowledge to provide the personalized information services and interactive interface.

Abstract:The resource flexible scheduling is the key technique in process simulation. After the introduction to the process engineering theory, the architecture of a new enterprise model was given out, which emphasizes the key of process simulating on how to control the activities according to the schedule polices under the support of the infrastructure system. It was proposed that use dynamic PERT/CPM technique to support the cooperate behavior control rules in project-oriented processes. Flexible simulation technology was introduced through the discussion about cooperate behaviors for process model to the improvement of infrastructure model, and activing technique which is used to support the hidden parallel process. The simulation algorithm was also introduced through the transforming of the five states of activity.

Abstract:In this paper, a quantifier elimination method for positive real theory with order by Volker Weispfenning is extended to a decision procedure, from which a new and finer decision method for eleminentary theory of real addition with order is given. It is proved that its subclasses of fixed number of quantifiers is in correspondant polynomial hierarchy. The result of this paper is essentially an extension of the claim on positive real theory by N. Megiddo to the whole real theory, which is relatively simply enduced in this paper, and comparably better than the similar result by E.D. Sontag.

Abstract:A distributed cache system in distributed information access environment is discussed in this paper, which is used to increase the access efficiency and reduce the communication throughout. The cache node can store the data from multiple data sources. Different access of information can reuse the same cached data and thus improve the hit rates. The cache system adopts various methods to keep the consistency of cached data. Finally, concept of virtual cache node is proposed, which extends the original caching system. The adoption of virtual nodes is for reducing the amount of replicatedcached data and reduce the computation cost for a global optimized accessing plan computation.

Abstract:Stereo matching has long been one of the central research problems in computer vision. A stereo matching algorithm is proposed based on gray level segments in order to recover dense depth map for the application of view synthesis in IBR (image-based rendering). This algorithm uses gray level segments as elements to be matched, and introduces the principle of DTW (dynamic time warping), which applied for speech recognition, into the matching processes on the segments. Covering more image space than point elements, gray-level segments are less sensitive to noise and photometric variations and easier to be matched, and can reduce the false matching. Moreover, gray-level segments segments require less computational resources than feature-based line segments and conices. Experimental results show that this algorithm is effective and of great value to use.