Abstract:A layered framework and related standards based on the analysis of the connotation, functions and open characteristics of PDMS(product data management system) are proposed in this paper. Based on the framework, three implementation methods are discussed: method based on the network database, method based on the groupware platform and method based on the Intranet and the Web technology.

Abstract:Qualitative reasoning can predict the system behavior of complicated system with incompleted knowledge, but the combinatorial explosion of reasoning branches limits its application. Traditional algorithms not only produce intractable reasoning branches, but also occupy a lot of memory space. It makes the result of reasoning difficult to be understood, and sometimes it even leads to failure. In this paper, on the base of LSIM, a method of qualitative reasoning based on LCQR(layered causal qualitative reasoning) is proposed to solve this problem. LCQR extracts the causal relation between variables, layers it and utilizes it in qualitative reasoning. The result of LCQR is satisfying.

Abstract:According to multimedia data model based on temporal logic and XYZ system, the authors designed and implemented a new multimedia authoring language MAL. This language supports the specification of abstraction, concurrency and the procedure of stepwise refinement design, it can be used to specify modeling information units, design and script a multimedia title in a unified framework. A visual multimedia authoring environment based on MAL has been developed.

Abstract:A data model named OIM is proposed in this paper as the common data model for integration of heterogeneous data sources. Based on rooted connected directed graph both cyclic and acyclic, OIM can describe the relationship of a complex object and its component objects, as well as the links between HTML files in World-Wide-Web naturally. As the metadata is associated with each object, OIM is especially suitable for describing objects without explicit predictable data schemata. As the formal foundation for query decomposition and optimization, a new algebra called OIM algebra, which includes six operations, i.e., object union, difference, select, project, paste and cut, is proposed. In comparison with relational algebra, OIM algebra is more flexible and powerful.

Abstract:Some navigation mechanism are introduced and analyzed. Based on PFNET(pathfinder net) which is a network model, according to nodes and links and users' custom, two types of navigation graph are presented to help the users browse information and solve the lost problem in hypermedia systems.

Abstract:SAQ is an experimental system to perform acquisition, verification and reusing of formal specification, in which the lexical and syntactic definitions of one concept should be integrated into one context-free grammar. If employed conventional context-free grammars to describe the overall definitions of complicated concepts such as natural languages and programming languages, separators such as spaces and carrier returns should be included and the definitions should be very messy. To solve this problem, a special kind of context-free grammars is presented. The grammars are obtained by dividing the set of non-terminals and the set of terminals of conventional context-free grammars into two respectively. As a result, the grammatical definitions of complicated concepts are relatively neat; at the same time, lexical analysis and syntax analysis can be integrated into one parsing process. In addition, the authors present the corresponding parsing and derivation tree construction algorithms, which are obtained on the basis of the general parsing method of Earley and its corresponding algorithm of construction of rightmost derivation respectively.

Abstract:FSS(feature subset selection) is an important problem in the fields of machine learning and pattern recognition. Minimum FSS problem has been proved NP hard. However, existing heuristic algorithms are based on the consistency of positive and negative examples set, and a more optimal feature subset is hard to be produced under the noisy data in application to real-world domains. In this paper, from the degree of statistics, the effects of noisy data on FSS is analyzed firstly, and a concept of consistent feature subset which contains error rate is given. Then a heuristic algorithm——EFS (entropy based feature subset selection) based on information-theoretic entropy measure and Laplace error rate is presented. It is also applied to two real-world domains and is compared with GFS (greedy feature subset selection). The experimental results show that EFS can produce more representative feature subset, and can solve the noisy problem in the practical application effectively.

Abstract:In this paper, the definition of inter-task secondary parallelism and its existence condition are presented. Communications between two tasks and the calculation of communication waiting overheads are discussed along with the exploitation procedure of task secondary parallelism. And the effects of code motion and tasks fusion for enhancing task parallelism and reducing communication overheads are also illustrated.

Abstract:A model of evaluating cost by number of nodes searched in solving binary constraint satisfaction problems is introduced in this paper. It has been applied to the random binary constraint satisfaction problems to analyze the phase transition property of solving cost. The theoretical cost induced from the model is compared with those practical costs evaluated by number of nodes searched, number of constraints checked and time consumed respectively. Based on this model, the effect of using solving heuristics to reduce cost is discussed and a new variable ordering heuristic is given.

Abstract:In this paper, the authors propose a general approach to selecting materialized views, and define a structure named BC graph(benefit-cost graph) as the basis for selection of materialized views. A algorithm based on BC graph is then presented for such selection. Finally, it extends BC graph to BC^＋ graph to take account of indices associated with views.

Abstract:In this paper, task tree is introduced to model a parallel functional program on distributed and message passing platform. Based on this model, the main issues of task partition, such as task size and parallelism, are analyzed with speedup as the performance measure. An optimized partition algorithm is given, and an implemented system——PARLisp's performance data is supplied as an example.

Abstract:Data mining is recognized as an efficient method for solving Data_Explosion and Data Rich and Information Poor. Mining association rules is one of the important aspects of data mining. The mining algorithm AR_SET for multiple_level association rules is given in this paper. The algorithm calculates the frequent itemsets by set operation——Union and Disjoin, and speeds up the mining. The experiments show that AR_SET is efficient. Also, some variants of AR_SET are discussed.

Abstract:In this paper, the preceding relationship between the checkpointing intervals is defined, and the “reverse-precedence” phenomenon is proved to be the essential cause of Domino effect, and three main resolving strategies for Domino effect are presented which are off-line or on-line Domino avoidance, on-line Domino detection and elimination, and on-line Domino tolerance.

Abstract:In this paper, the authors discuss how to solve a set of typical constraint satisfaction problems, Job-Shop scheduling problems. Based upon the depth-first search, formal strategies of enforcing consistency, selecting operation and selecting start time are given, heuristic strategies of enforcing consistency and incomplete back jumping are introduced to further enhance the efficiency in solving Job-Shop scheduling problems.