Abstract:Rapid prototyping as a new method for software development is actracting great attention．The XYZ system is a CASE environment based on temporal logic and con forming to various ways of programming．The method for software development supported by the XYZ system is that of“formally specifying the target system，and then verifying or rapid prototyping the specification in a way of stepwise refinement till the effectiveiy exe cutable programs are obtained”．This method not only can make every steps in the evolu-tion procedures froom abstract specification to effectiveiy executable programs natural and eiear，but also can evaluate the correctness of the results of refinement in order to discover the problems and resolve them eariler．In this paper，the authors present and discuss the rapid prototyping which will be applled to the software development with the support of the XYZ system and the software tools——XYZ/PROT with which they support the rapid prototyping by construction the PROLOG prototypes from the specifications in temporal logic and executing the prototypes．

Abstract:Two most influencial data structures for modeling i．e．，winged edge and ra-dial edge structure are analysed．A new data structure——structure cycling between adja-cent dimensions is proposed．Similar to radial edge structure，the new structure is capable of representing wireframe，surface and solid models in a unified scheme．It is also capable of representing a model with some vertices having arbitrary number of adjacent edges and some edges having arbitrary number of adjacent faces．The topological part of the new structure is about half in size compared with the same part of the radial edge structure．However the efficiency for accessing topological relation is the same for the two structures．

Abstract:The observable computing actions of a Turing machine，including moving and writing，can be categorized as the behavior of the Turing machine．In this paper，it is shown that the behavior of a Turing machine is learnable．That is，there exists a learning procedure which，given the behavior of a Turing machine，will yield a new Turing machine with the same behavior in a finite number of steps．According to the same mechanism，the behavior of an LR parser is also learnable．

Abstract:In this paper，the authors present a method for description of graphics constraints，recognition of constraint and self—organization of automatic match between dimension and drawings．In this way，multi—direction drive among dimensions part drawings and assembly drawing is established．By a series tests．it is indicated that this method is efficient，practical and general．

Abstract:Based on the object—oriented database system FOOD which has been developed by Fudan University，a deductive query language O—Datalog is presented in this paper．O—Datalog language can easily express various requirements to query and deduce from the object—oriented database．It can be translated to a Datalog—like form，SO the O—Datalog programs can be evaluated using the efficient algorithms for evaluating Datalog programs．It'S easier to implement the language than to implement those based on non—Horn logics．O—Datalog is firmly an extension of Datalog in form．The syntax and the semantics of the O—Datalog language are discussed in this paper．

Abstract:The last paper has introduced the O—Datalog，a deductive query language of the object—oriented database system FOOD．This paper puts great emphasis on dis-cussing some transformations of the O—Datalog programs，and proving that these trans-tormations preserve the same semantics．So an O—Datalog program can be transformed into a corresponding Datalog program and can be evaluated using the Datalog program，then an algorithm of evaluating the O—Datalog programs is given．

Abstract:This paper presents a formal conceptual model for multimedia objects，name-ly，OMCOM(object—oriented multimedia conceptual model)．First，the particularities of the multimedia data modelling are analyzed，the inadequacies of existing multimedia data models are discussed．As a result，the idea is proposed that multimedia data modelling should be divided into two layers：one is the conceptual layer which models the contents and structures of multimedia objects，and the other is the presentation layer which models their presentations including their temporal and spatial relationships．OMCOM is the roodel used to implement the conceptual modelling．Secondly．the formal definitions of the elements in OMCOM are given according to the requirements of multimedia conceptual mod-elling．Finally，object algebra is defined，which is based on the lattice property of OM-COM instances．Object algebra is the system of operations over multimedia objects on the conceptual layer．The formal definitions of object algebra operators and expressions are given．

Abstract:This paper studies the design and analysis of parallel FFT algorithm on three parallel computational models including APRAM，BSP and LogP models；reveals the in-trinsic properties and relationship between these computational models；comments on the utilizable and operationable properties of them．

Abstract:This paper introduced all object—oriented graphics support system．The authors used object—oriented concept to design this system，so that one can design graphical application model at high level of abstraction and translate his meaning into a model maturely．The system is designed to be easily extensible and reuse．By adding some special knowledge about some area，the user can tailor the system into a more special one．

Abstract:The ODP(open distributed processing)aims to solve the software Interface problems to achieve portability，interoperatability，and distribution transparency．This paper presents some of the existing problems for ODP to achieve its goal, analyses the ser vice static binding mechanism of the traditional RPC model in the view of ODP．An im-proved，supporting service dynamic binding RPC model is proposed，through which better performanee can be obtained．

Abstract:X.25 SCB(synchronous communication board)iS the efficient method for PC users to access PSDN(packet switching data network)．Presently，X25—SCBs are imple-mented with intelligence，high—speed and high—volume storage．The thesis discussed the design and implementation of a high—performance synchronous communication board TH—X25．Its architecture，service primitive and the design and implementation of its driver are given．

Abstract:In the progress of draft automatic inputting in computerized embroidery punching system，color clustering is a suitable method．It includes two ways，K—averag-ing iteration and combination．In contrast to traditional edge detection，these two ways are more effective．They have been applied to the image preprocessing system for computer—controlled embroidery．

Abstract:Distributed graphics is the study on graphics processing in distributed envi-ronments，which is the product of integration of computer graphics and distributed sys-tern．Based the authors’research works，they give a brief overview of researching areas of distributed graphics．Then they discuss on the applications of distributed graphics in ani-mation,ViSC,multimedia and virtual environment．Finally they have a forward 1ook at the development of distributed graphics．

Abstract:This paper describes a program understanding support system for C++lan-guage：HOOPUS(hypertext—based object—oriented program understanding support sys-tem)．This system is based on program structure analysis technique，it shows the program information in a clear way to facilitate understanding the function and structure of C++program．It also uses hypertext，multi—dimension view and navigation technique to facili-tate program understanding process．

Abstract:The design of authentication protocols is notoriously error—prone．ISO has been working on a variety of authentication protocols standards for some years．This paper explores the approaches to formal analysis of authentication protocols using BAN logic and points out that the BAN logic analysis does not always lead to correct outcome．Finally，this paper discusses the design principles of authentication protocols and the future direc-tions for improvements of the BAN logic．

Abstract:This paper introduces two solutions using STEP to integrate various CAD/CAPP/CAM application systems and presents the implementation methods of the EXPRESS compiler and the STEP exchange file pre/post processor as components of GHModeling4.O System．Some essential techniques and strategies such as SDAI data dic-tionary structure，schema manipulations，the STEP exchange file accessing and the appli-cation protocol conformance are discussed in detail．

Abstract:Virtual environment ia an advanced human—computer interface which use multiple media to generate human'S visual and auditory effects．A concept of overall repre-senting temporal and spatial information of sound was provided in this paper．Infinite loudspeakers were addressed in 3D space to represent temporal reality and spatial reality of a sound from any spatial direction．With discussing the essence of spatial sound，a realistic spatial sound rendering algorithm based on 3D vector analysis and a system implements-tion based on time—critical computation were provided．Comparing with traditional three —dimension sound synthesis method based on HRTF,this method required less computa-tion to generate more realistic sounds，

Abstract:The task assignment of AGV is a sophisticated combinational optimal prob-lem．It is NP complete．A new method，which is different from the general searching from task node to task sequence，is given in this paper．The new planning process goes from key route to the clustering based on key route then to task sequence．So the method is called“cloud room effect”algorithms．As a satisfied solution is given．the ability of real time of the algorithms is improved．

Abstract:Term rewriting systems are a general model of computation．The normaliza-tion property of a reduction strategy for TRS's plays an important role in the application of TRS's，and overlapping rules and other types of ambiguities complicates reduction strate-gies greatly．This paper deals with these ambiguities in general and proposes art approach to a general class of problems involving ambiguities．

Abstract:POLYBASE is an integrated multi database system．This paper mainly dis-cusses the architecture and integration data model of the POLYBASE system，presents a query presentation model based on the concept of weighted query trees，designs and imple-ments an effective navigative query processing algorithm on the basis of the concept of weighted trees．

Abstract:Using hypertext style and object—oriented technology，the authors proposed the model WNCH(works，node，content，hotspot)for the management and organizing of multimedia works．This project is based on the unified view of readers and authors and the properties of multimedia works．In this paper，the properties of multimedia is analyzed，some examples of multimedia information models are introduced，the structure，design and expressing ability of WNCH model is described．At last，the application of WNCH model to multimedia works management including authoring，reading and guiding is discussed．

Abstract:To overcome the notation redundancy in the modaI resolution systems estab-lished bv P．Enjalbert and L．Farinas del Cerro，the authors propose marked modal resolu-tion．whose soundness and completeness are proved．In marked modal resolution，they mark possible operators in modal clauses，and add a rule for computing resolvent of two formulas which are bound by possible operator respectively．This new method has the fol-lowing feature：a resolvent may not be a logical consequence of its parents，it is only a log-ical consequence of the input set of clauses,that iS the soundness．At the same time，they design and implement system RD based on modal resolution of Enjalbert and Farinas del Cerro．and system MRD based on marked modal resolution in C-PROLOG，and then run on SUN workstation．The experimental results show：the MRD method iS almostly 10 times faster than RD．

Abstract:This paper has defined the formal semantics of object identities，values，query predicates and database states for complex object databases with a hierarchy of alge-braic specifications．Semantics of queries，especially recursive queries on individual objectsis presented based on insights into query paths．Finally，the semantics of selection opera tions on OODBs is given．

Abstract:This tool is based on a graphic language called XYZ/CFC which is the graph-ic representation of a temporal logic language XYZ/E．By means of this tool，a XYZ/E program of unstructured form can be transformed into structured form．XYZ/E iS suitable to play the role of an intermediate language．As the comequence，this tool can also be used to transform a program of any conventional language into its structured form．

Abstract:The authors introduced the concept of U—cyclic in this paper，and it is proved that all U—cyclical terms are easy terms．So that the reductiveness on a class of easy teams is given．Moreover this property is important to researching halt problem．

Abstract:In this paper a new scheme of key management is provided，two—level key transformation，which is considerably secure and easy to be implemented．Based on this scheme，a DBEMT(database encryption management tool)is achieved．The overall de-signs and functions of DBEMT are presented in detail．according to various ways for secu-rity class division，the designs of cipher databases using field classification,record classifi-cation and attribute classification methods are discussed respectively．Compared with conventional schemes，it has properties of good operability and high running speed．This scheme offers good prospect for utilization．

Abstract:A new kind of neural networks for continuously computing the shortest path problem is presented in this paper，and the stability of the network is proved．This kind of neural networks is composed of the minimum selecting networks and other usually used neurons．For any directed or undirected graph，the neural networks are always convergent to their single equilibrium,so that the shortest paths of all the node pairs are produced．This paper makes a new study on solving the optimization,but not NP—hard problems by neural networks．

Abstract:This paper developed XYZ/NGAE system based on AG(attributed grammar)．The system generates the FEC(front—end compiler)of a programming lan-guage automatically．In this paper，AG and the language for attribute definition are intro-duced at first．then the system structure of XYZ/NGAE and the attribute optimization technique are described．

Abstract:This paper spreads the lock resolution and the generalized lock resolution in the first-order logic to the BOFL(Boolean operator Fuzzy logic)．It is shown that the lock resolution and a special generalized lock resolution in BOFL are complete in the broad sense．

Abstract:In the context of processing the multimedia information accompanying with the historical events，a lot of events can be viewed as growth procedure．This paper gives a mathematical model for the growth phenomena，and discusses its additive property，monotonicity and stability．It also introduces the ignore_trivial algorithms and non—triv-ial increment algorithms in MHBase——the prototype of multimedia database system．

Abstract:How to describe the dynamic process of belief functions in the Dempster—Sharer theory is a very important research field．The specialization of evidence functions is a kind of important method to express the dynamic process．In this paper，the authors dis-CUSS two different definitions for the specialization，correct Kruse'S mistakes in his book“Uncertainty and Vagueness in Knowledge Based Systems”and prove theorems 6.10 and 6.11 in the book by using adverse orders．At last，based specialization，the authors discuss some relationships among three kinds of condition rules which lead to dynamic pro-CeSS of belief functions．

Abstract:Based on the ELF(Edinburgh logical framework)and Martin—Lof logical framework，the authors propose a logical framework that unifies the two languages．Their logical framework is well suited for semantic analysis．Examples are given to show how to encode object languages in the calculus．

Abstract:Function programming language with iteration operator is an efficient，useful and practical，high—level programming language with sound theoretical foundation．It can represent recursion and iteration．This paper proved that partial iteration functions which are defined by iteration operators from given iteration functions can compute partial recur-sion functions．It means that any turing computable partial functions can be c6mputed effi-ciently by iteration functions with sound theoretical foundation．In this paper,the authors defined the composition operator，primitive iteration operator，minimization iteration oper-ator and partial iteration functions．Then they proved that partial reeursion functions can be computed by partial iteration functions．

Abstract:A new computation model of programming language，function model with iteration operator was presented．The syntax of language based on the model，the set of computation rules and a proof of determinism were given．In this language，sequence is expressed by composition of expressions，branch by condition expressions and iteration by iterative expressions．such that its programs are based on mathematical semantic founda-tion，easy to understand，prove its correctness and constitute efficient execution system． The authors believe that this language is an efficient，useful and practical，high—level de- terministic programming language with sound theoretical foundation．

Abstract:A parallel quicksort algorithm is given and which，given an EREW PRAM with k processors，sorts n items in expected O((n/k+logn)logn)time．and thus the prod-uct of time and number of processors is O(nlogn)on the average for any value ofk≤n/(logn)．

Abstract:LNDBMS is a data collection oriented DBMS with high performance，which uses CLIENT/SERVER architecture and supports MULTI—SERVER．It is also an open system，by which users can manipulate transparently multiple heterogeneous databases．LNDBMS has been well applied to a large project with multiple servers．Its design and implementation are described．

Abstract:Communication overhead in massively paralleI processing systems is an im-portant factor which affects the performance of MPP．Broadcasting is a communication method used frequently in MPP systems．If broadcasting can be implemented as quickly as possible，then the performance of system will be increased．Based on the TORUS inter-connection network，this paper proposes the pipelining broadcasting，analyses its perfor mance and shows that pipeline broadcasting can derease the broadcasting delay and im- prove the performance of MPP systems．

Abstract:Based on the idea that the domains of attraction of all memorizad patterns must possess balanced shape,a minimax criterion for design of HoFfield associative memo-ry，which requires the smallest domain of attraction to be maximized，is proposed in this paper．A quick learning algorithm is first given，and furtherly，a constrained perception optimization algorithm is developed．A large number of simulation results confirm advan-tages of these algorithms given in this paper over existing ones．

Abstract:This paper addresses a Pet ri—net based approach to supporting specification construction and execution of software system．On the basis of requirement descriptive model，a kind of specification represented in Petri一net is formed，which can generate the functional model on the implementation level by applying transformation and refinement．The specification is validated and verified by different simulation mechanism．This paper has put forward a kind of specificeation framework and may provide a novel idea for soft-ware automatic generation from the specification Ievel．

Abstract:One—way hash functions is an important part of cryptography.In this pa-per,a new method for constructing one—way hash function F from any fixed size one—way hash function f：∑^m→∑^t,m>t，is presented，the property of which is that it is easy to be parallelized．

Abstract:Based on the fundamental idea of RSA，the authors designed a dynamic pass-word authentication scheme，solving the problem in the Lamport scheme．In this scheme，there is no requirement that every user should hold a public key cryptosystem or a signa-ture system，thus its system complexity is rather low．A slight modification on the funda-mental scheme can also lead to an identity—based dynamic password authentication scheme．

Abstract:DSM(SVM)is a memory model which was proposed at the 1980s'．DSM provided a possible way to solve the problem that scalability was contradiet with easy pro-gramming．Because DSM was built on the top of message passing scheme．and the cost gap between local reference and remote reference is quite large，the efficiency of DSM will be very low if the authors simply adopt the cache coherence semantic for multiprocessors．So，some Weaken，relaxed coherence semantic model have been proposed,but these mod-els are quite different from sequential Coherence which was assumed by most programmers intuitively，so programming and debugging becomes very difficult．This paper proposes SWHC(strict weak hybrid coherence)model which can eliminate false sharing and provide more temporal and spatial locality,also provides a scheme to keep Seqhential coherent．Experiment results prdved that strict and weak hybrid consistency semantic can eliminate the“thrashing”and improve system performance greatly．

Abstract:Object—oriented concepts provide some useful methods for designing s rela-tional database．In this paper，the concept of inheritance constraints is given．The authors show that the inference rules for inheritance dependencies are sound and complete．Fur-thermore,a series of results about the closure of inheritance dependencies，the minimal set of inheritance dependence is obtained，which arc analogous to functional dependencies．

Abstract:This paper describes the design and implementation of an object—oriented software automation system JDAUTO/0．JOOSL is an object—oriented software formal specification language．Based on JOOSL，the authors have developed the PDAUTO tool，a semiautomatic tool which synthesizes detailed designs from preliminary designs，and the DDAUTO tool，an automatic tool which transforms detaihd designs to C++programs．

Abstract:This paper proposes a fast convergent genetic algorithm to overcome the slow convergent rate of conventional genetic algorithms，and takes 8一queen problem as an example to examine the algorithm．The simulated results demonstrates that the algorithm'S convergent rate is faster than that of the conventional techniques．

Abstract:Phrase boundary location provides an important information for bracketing and tagging the phrase automatically．This paper describes an experimental model for the automatic prediction of the phrase boundary location．It consists of three processing stages：first，automatically identify the phrase boundaries using statistics from treebank；then，post—tune the results using local tuning rules generated by an error—driven the machine learning method；at last，refine the results of the last two stages with the overall tuning rules summarized by man．Experimental results on a corpus of 1 434 sentences demonstrate a high rate of the success for predicting the phrase boundary(96.33％correct the prediction for the close testing and 94.54％correct the prediction for open testing)．

Abstract:An operational profile describes how users employ a software system．By developing operational profiles of a software system，operator can guide testing efficiently and increase software reliability as maximum as possible．This paper mainly analyses and discusses basic processes and main methods of developing an operational profile，on the basis of introducing main concepts in operational profiles．

Abstract:Parallel—outermost strategy is widely used in reduction systems．Parallel—outermost pattern matching studies the pattern matching method adapted to the strate-gy．By preproeessing the patterns into an adaptive pattern matching automaton APMA，this paper takes advantages of local matching information of unsuccessful matching states to build parallel—outermost pattern matching automaton POPMA．By use of POPMA，all of parallel—outermost pattern matching subterms can be gotten through a single scan．The method limits the size of the automaton，avoids the overheads in the naive algorithm and makes the good relation between the pattern matching and the reduction strategy．POPMA also can be used to improve parallel—outermost strategy．

Abstract:Programming language based on state logic is an efficient，useful and practi-cal，new pure logical programming language with sound theoretical foundation．It can represent recursion and iteration．With reeursion its computability can be provided．This pa-per proves that partial recursive functions can be computed only with iteration programs in state logic．It means that any computable object in Turing machine can be defined itera tively and computed efficiently with Pure logic．

Abstract:Operator programming language is an efficient，useful and practical，high—level，new programming language with sound theoretical foundation．How to compile this language is an important problem which should be made researches on．The translation ot programs from procedure language into machine language has been solved，hence the linchpin upon which the translation depends is the transformation of operator programs to procedures．In this paper．the transformation semantics of operator language to while pro-grams was given with model match method．A basic set of transformation rules was given and it was proved to be complete and consistent，so that the theoretic and technologic foundation to translate operator language to procedure language was provided．

Abstract:This paper discusses two approaches to the specification of concurrent sys- terns：temporal logic approach and state machine approach．As a result of this discussion， the authors propose a new kind of formalisms of specification：fair transition system speci- fication(FTSS)．This specification approach combines the best features of temporal logic and state machine methods and revises these drawbacks that temporal logic approach USU-ally is complicated and not easy to understand，especially，it fails to be used for "local"properties of concurrent systems．The authors further consider each component of FTSS and get conclusions that FTSS is machine closed and the specification process is consistent and complete．An example of a lossy—transmission protocol shows that the presented ap-proach is simple and easy to understand and to use．At the end of the paper，some applica-tions of FTSS are given．The approach provides a unified framework for program verifics-tion and step—wise refinement of concurrent systems．It has been successfully applied to program verification．

Abstract:This paper presents an algorithm to find an optimal cover of the logic node'S functions in multilevel logic structure，and proves the efficiency of the algorithm in opti-mizing the multilevel logic structure．

Abstract:Feature modelling has been received wide respect in recent years，and is becoming the efficient method to achieve the new generation of CAD．This paper firstly gives the formal defenition of feature，then offers a form feature based respresentations of objects，at last provides an important model in the applied modelling system，namely form feature modelling．

Abstract:The properties of the secondary fixed—points inλ—calculus are discussed in this paper．The author first shows the relationship among three propositions for the sec-ondary fixed—points and proves these propositions，and then formulates a sufficient condi-tion for the so—called secondary fixed—point combinators and gives a series of such corn-binators．At last．the author establishes the second fixed—point theorem in the n—ary version．

Abstract:After pointing out the main shortcomings of VDM in developing larger pro-grams，this paper introduces a new concept—DD(date decomposition)which is based on the ideas of model split，modularisation and operation decomposition，and combines it with VDM to form a more general formal development method DD—VDM．As a result，a more flexible development strategy can be adopted and the development complexity can be effectively controlled．

Abstract:This paper first proposes a data parallel algorithm and a replicated data algo-rithm for the graph theoretic operation of computing the single source shortest paths and demonstrates the generality of the data replicated technique by applying it to the graph theoretic operation for solving SSSP problems，then computes and discusses the speedup of the replicated data algorithm over the data parallel algorithm，finally points out that the data replication technique is not only applicable to fast image analysis，hut also is general enouth to be applicable to a large class of the graph theoretic operation problems．

Abstract:This paper presents a 2—phase schema transformation method for schema transformation from relational databases into object—oriented semantic databases．The method is based on the semantics identmcation．The semantics included in local relation schemas can be divided into two parts：one are intra—relation semantics existing within local relations and the others are inter— relation semantics existing among local relations．All intra—relation semantics are captured into the global integration schema in phase l，and all inter—r elation semantics are enriched into the global integration schema by schema reconstruction operations in phase 2．The mapping rules and transformation operations are defined for setting up the transformation mechanism，and a set of algorithms and transfor mation formulations are provided for generating G—relations．

Abstract:An algorithm for outline resolution on 2D complicated PCB circuit and irregu-lar pattern is presented．Its mathematic model，algorithm description and implementation are described as well as correctness proof and complexity analysis of the algorithm．The sample results on a PC386—based CAD/CAM system(CACAO)are also demonstrated．

Abstract:In this paper,the authors address the problem of biotical vessel segment in the vessel images taken from human body'S burned skin．These blood vessels have poor lo-cat contrast with multi—diameter，and take on network distribution．The authors intro-duee an algorithm based on the optical and Spatial properties of objects to he recognized．Through sefinning the vessel image in various angles．anaiysing the glay—level profile of the cross section of blood vessels and detaching vessel image according．to vessel diameter，they could obtain the fixed quantity distribution of biotical blood vessels．This segmenta-tion algorithm is efficient and automatic，it can help physicians in diagnosing diseases and show promise in many relative fields．

Abstract:If∑is a finite alphabet。w--language is the set which consists of some infi— nite strings over∑．e--regular language is the~o--language that be accepted by o--finte automata．The authors described a subclassω—language from set viewpoint．It cannot tradition that gives the subclass ω—regular language from generation or reeognizition．In this paper．the authors also show the subclass ω—regular language from set viewpoint．This subclass is more widely．

Abstract:Considering some AM(Associative Memories)(such as Kohonen and Murakami models)implemented with generalized inverse lack complete recall to the stored data and nonlinear mapping abilities，in this paper an improved associative memory model with complete recall and nonlinear mapping abilities is presented by adding one extended (hidden)layer to original AM models．The theoretical analysis based on matrix SVD(sin-gular value decomposition)and the simulation on computer show the performance superi-ority of the improved model．

Abstract:The software project estimation is an important part of the software project management．This paper is justly a resume of author's working experience in an European company.

Abstract:This paper presents a powerful notation for set comprehension in functional languages and a program transformation for its implementation．Introduction of the set notation not only increases the expressive power of functional programs，but also offers opportunity for improving their running efficiency．

Abstract:In this paper，the authors present a main memory database system ARTs—MMDBS，aod discuss with the emphasis on its database management mechanisms and ae-cess methods．And the paper describes in detail a new index Structure—SB tree used in ARTs—MMDBS．

Abstract:Augmenting PTL(propositional temporal logic)with a 2—partition operator,which is a natural generalization of unless operator，leads to a simple but complete frag-ment of Wolper—Vardi—Sistla's ETL(extend PTL)．It has succinct deductive system，better decision algorithm，easy translation to ETL，and is an ideal specification formalism for finite—state concurrent programs．

Abstract:This paper proposes rough number notion based on rough sets，and crests its operational laws．It illustrates its application in the practice．Finally，it describes theoreti-cal value of rough approximate operator λ in rough logic．

Abstract:This paper presents a multimedia intelligent database system for market anal-ysis and forecast applications．The system extracts data necessary for market analysis and forecast from management information systems，analyzes market situations，and forecasts sales using both mathematical and intelligent methods．The analysis and forecast results are summarized in a system—generated text report and then presented to users in a syn-thesized speech which is synchronized with the display of tables，statistical graphs，visual-ization models，and real—time 3D background models．After briefing the system require-ments and functions．this paper focuses on the system'S general architecture and overviews some of the key issues and solutions in its development．The paper concludes by indicating that a general intelligent database system framework may be built On the basis of the sys-tem presented．

Abstract:This paper presents a general framework and algorithms for a case—based combinatorial forecast method，which emphasizes the combining of knowledge ahd mathe-matics．By analyzing expermental results，the paper discusses the shortcomings of various methods of forecast．

Abstract:In this paper，a market analysis framework and its implementation are pre-sented．This framework considers market analysis as an information Process course to enhance the quality of making market decision．The paper discusses the structure 0f this market analysis framework and gives the preliminary research results of some kev issues such as trend analysis，qualitative causality analysis and etc．

Abstract:Market analysis and forecast system is a decision support system that pro-vides market information for decision makers of companies．In this paper，the architecture of a market analysis and forecast system is discussed，and a template based method of text report generation and its implementation in the system are presented．

Abstract:In this paper，a method for the market analysis based on the rules is present-ed and the expressing ability of the rules is discussed in conjunction with the features of random market analysis．The concept of S—expression is used to establish a unified store structure．The concept of variable environment is brought forward and a unified algorithm is realized．Finally，a rule—based report generation system for market analysis is build up.

Abstract:This paper is intended to explore solutions to the data requirements of a mar-ket analysis and forecast system though application of data warehouse technology．It pre-sents several aspects related to the building of a data warehouse，including data extraction and check—up，data normalization，and a mechanism to support market analysis and fore-cast．The paper also summarizes some basic requirements for applying data warehouse．

Abstract:A computer system can use multiple types of media to enrich the expresion of same piece of information．Thus，while presenting the information to end user，the system faces the problem of multimedia synchronization．This paper gives some exploration on the problem，and proposes a control model．The paper also gives a detailed implementation of the modeI as well．

Abstract:The real time display of 3D complex models，which is widely used in flight stimulation，scientific computing visualization and architecture walkthrough，is becoming one of the most challenging issues in Computer Graphics．In this paper．the authors have developed several efficient techniques to perform real time display，including multi—reso lution model，hierarchical model，instancing，visibility determination，texture mapping，levels switching and controlling and so on．These techniques have been successfully ap-plied to the information management and decision support system of Yuxi Cigarette Facto ry．Finally，the result is given and the future work is presented．

Abstract:In the management information system，it is an important field in current visualization techniques applications to display the complex management information in a vivid mode．In this paper，the authors try to signify the data and their relations with the different 3—D entities．Two visualization models．Pipe Model and Lath Model，are also presented in this paper．

Abstract:Prosody understand is the foundation of speak synthesis．After investigating the state of research in TTS．A method of prosody markup is proposed．Grammar and symbols of prosody markup are designed．The proposed method is used in a Chinese TTS system．As a result of it．simulation of stress and intonation is achieved．The naturalness of synthesized speech is improved．

Abstract:There are many new features in modern software development．New demands are needed in software testing，such as more efficient，user—oriented etc．Traditional testing methods are unable to solve these problems．In this paper，the authors describe a new testing strategy to meet these demands，which is called application operating coverage based software testing(AOC)The fundamental analysis tools of AOC are navigation dis gram and event record serials．At the end of this paper，they give an automatic testing tool for AOC．

Abstract:A quantity model of multi—agent computational market is presented．The authors define computational market，task，Consumer and Server entity，prove there ex-ists an equilibrium in the market，and use the distance from the optimal allocation of com-puting resources as the criteria of the market proficiency．

Abstract:This paper discusses a parallel programming tool based on the task assign ment and communication planning heuristic algorithms．It consists of four components：task allocator，communication planner，communication schedule，and communication statement generator．A program framework composed of communication statements can be generated and used at the programming phase．By using this tool，developing parallel pro-grams will become easier and it is certain that high efficient programs can be produced．

Abstract:In real world，agents are facing a continuously changing environment．They should remain reactive to emergencies，construct plans dynamically，model the world and other agents to predict future events，understand and generate language to communicate with other agents．How to organize all of these characteristics in an agent is an important issue．The paper proposesd a compound agent architecture．Each agent contains modules specialized in perception，action，reaction，planning，modeling，and communication．All modules run concurrently，and are coordinated by a decision making module．Through in teractions of these modules，the agent is capable of performing more rational behavior in real world．Using MAPE，which is a software environment proposed and implemented，it is quite easy to construct agent with compound architecture．The authors also proposed a new modal logic semantics for multi—agent modeling and reasoning．The end of this paper will compare the proposed architecture with several other agent architectures．

Abstract:KBASE—P is a knowledge—base programming language．It uses KBASE as query language and FD—PROLOG as its procedural host language．The two languages have the same programming paradigm—Horn clause paradigm，and the KBASE—P system accesses disk—resident data in set—at—a—time mode．thus the impedance mismatch problem has been minimized in KBASE—P．KBASE—P manages facts and rules including intermediate results with the help of RDBMS，hence it can manage large amount of data which can be shared by users，and is suitable for large—scale applications．KBASE —P language and its features are described in the paper，and some comparison with related systems are given also．

Abstract:Destructor Eliminating，term generalizing and irrelevance eliminating，fertil-izing are several important inference strategies in induction inference system．Starting from the introduction of the above strategies，this paper discusses the basic idea of the strategies and the method of using the strategies，The system has been implemented by using compiler LISP on micro computers．

Abstract:This paper discusses the clause simplify strategy of induction inference sys-tem．The inference ability of the system mainly depends on the ability of simplifying the clauses．Starting from the type prescript of the definition，this paper introduces how to compute and use the type set information to simplify the clauses，how to use the rewrite strategy to simplify the clauses．The system has been implemented by using compiler LISP on micro computer．

Abstract:This paper discusses the term rewriting strategy in induction inference sys-tem．Axioms，rewriting lemmas，function definitions，term rewriting rules are used to rewrite the term to be rewritten．The system will get a rewritten term which approaches to the inference goal．This strategy has been implemented by using compiler LISP on micro computer.