It is often necessary to combine multiple scores into a joint decision in many real-world applications. To that end, an essential challenge is to build a proper weighted scoring rule, which assigns more weight to the more important scores and derives

Conventional k-means only considers pair wise similarity during cluster assignment, which aims to minimizing the distance of points to their nearest cluster centroids. In high dimensional space like document datasets, however, two points may be nearest neighbors without belonging to the same class. Thus pair wise similarity alone is often insufficient for class prediction in such space. To that end, in this paper, we propose to augment k-means with pair wise constraints generated from group similarity-based hyper clique patterns, which consist of strongly affiliated objects and serve as more reliable seeds for classification. Experiments with real-world datasets show that, with such constraints from quality hyper clique patterns, we can improve the clustering results in terms of various external criteria. Also, our experiments indicate that even if few constraints are violated in the original result of k-means, imposing many quality constraints may still bring gain of performance.

Data privacy refers to the issue of how to preserve the confidential information in individual data cells while still being able to provide an accurate estimation of the original summation values for range queries. There are three major goals in data privacy:l)Security - the data must be protected from being revealed; 2)Accuracy - results of analysis are valuable in a business's decision making and 3)Accessibility - data should be as easy to be accessed as possible. However, the open nature of data warehouses creates a security conflict. Another conflict is between security and accuracy of the query results. With the increase in the number of data warehouses and OLAP users, the misuse of data warehouse is steadily growing, which has led to the needs for proper techniques to support all three goals of data privacy. We propose a novel and simple but effective solution to fulfill all three goals.

An important issue that needs to be addressed when using data mining tools is the validity of the rules outside of the data set from which they are generated. Rules are typically derived from the patterns in a particular data set. When a new situation occurs, the change in the set of rules obtained from the new data set could be significant. In this paper, we provide a novel model for understanding how the differences between two situations affect the changes of the rules, based on the concept of fine partitioned groups that we call caucuses. Using this model, we provide a simple technique called combination data set, to get a good estimate of the set of rules for a new situation. Our approach works independently of the core mining process and it can be easily implemented with all variations of rule mining techniques. Through experiments with real-life and synthetic data sets, we show the effectiveness of our technique in finding the correct set of rules under different situations.

Document image processing begins at the OCR phase with the difficulty of automatic document analysis and understanding. Most existing systems only do well in their specific application domains. In this paper, we describe a domain-independent automatic document image understanding system with learning ability. A segmentation method based on "logical closeness" is proposed. A novel and natural representation of document layout structure-a directed weight graph (DWG)-is described. To classify a given document, a string representation matching algorithm is applied first, instead of comparing all the sample graphs. A frame template and a document type hierarchy (DTH) are used to represent the document's logical structure and the hierarchical relationships among these frame templates, respectively. In this paper, two learning methodologies are applied-learning from experience and an enhanced perceptron learning algorithm.

A knowledge-based tutoring system (KBTS) is a computer-based instructional system that uses artificial intelligence techniques to help people learn some subjects. We found that the knowledge communication process involving a KBTS and a human student can be decomposed into a series of communication cycles, where each cycle concentrates on one topic and contains four major phases: planning, discussing, evaluating and remedying. The major contributions of this work are the development of a generic architecture for supporting the knowledge communication between a KBTS and a student, and a graphical notation and schema for supporting the curriculum knowledge representation and manipulation during the planning phase of a tutoring process. The curriculum knowledge about a course can help a tutoring system determine the sequences in which the topics will be discussed with the students effectively and diagnose the students' mistakes. The curriculum knowledge base contains the goal structure of the course, prerequisite relations, and multiple ways of organizing topics, among others. As an example, we focus on developing SQL-TUTOR, a KBTS for the domain of SQL programming. This system has features such as an efficient control mechanism, explicit curriculum knowledge representation, and individualized private tutoring. For allowing the students relative freedom to decide how to study the domain knowledge about a subject, the system provides the students with a group of operators to hand-tailor the learning schedules according to their special backgrounds, requests, and interests.

This paper presents a fuzzy-logic approach for analyzing typographical structures of textual blocks in order to be used for character preclassification. An efficient baseline detection method embedded with tolerance analysis is developed for locating precisely the baseline. Fuzzy logic is taken into account when the decision ambiguity of typographical categorization is occurred. The constraints on the fuzzy membership functions are formulated. Their boundary conditions are considered to preserve the continuity. An improved character recognition rate can be achieved by means of the typographical categorization.>

This paper presents the design of a knowledge-based tutoring system (KBTS) for teaching students to write SQL programs. After analyzing the underlying control mechanism in a tutoring process, we propose a novel architecture to support the process. Our system has many features such as the use of a uniform control flow strategy, the employment of a graph-based scheme to represent global knowledge, and the allowing of multiple teaching sequences and multiple viewpoints for a teaching goal. We argue that these features are essential for not only SQL tutoring systems, but also KBTSs in other domains. Thus, the paper establishes a generic framework for developing the various KBTSs. We also report some implementation considerations for the proposed system.>

Office information systems (OISs) are employed to support office workers in their management of information and to assist them in their daily work. In the OISs, document classification is one of the major functional capabilities. Classifying a document can be facilitated through the layout analysis of the document. A new approach to the layout analysis, called nested segmentation, is introduced. The layout relationships of components of a document are defined in terms of the adjacency of blocks. Given the adjacency of blocks, an adjacent block graph is introduced where the problem of the nested segmentation is transformed to a classic minimal cut problem for the graph. Also, an ordered labeled tree structure (L-S-Tree) is introduced to represent the segmented document for document classification.>

The authors present the design of a tool for classifying office documents. They represent a document's layout structure using an ordered labeled tree, called the layout structure tree (L-S-tree), based on a nested segmentation procedure. The tool uses a sample-based approach for learning, where concepts are learned by retaining samples and new documents are classified by matching their L-S-trees with samples. The matching process involves both computing the edit distance between two trees using a previously developed pattern matching toolkit, and calculating the degree of conceptual closeness between the documents and samples. The experimental results show that the tool is capable of classifying various types of office documents, even with very few samples in the sample base.

This paper presents the design of an intelligent document processing system, called TEXPROS. The system is a combination of filing and retrieval systems, which supports storing, extracting, classifying, categorizing, retrieving and browsing information from a variety of documents. TEXPROS is built based on object-oriented programming and rule-based specification techniques. In this paper, we describe main design goals of the system, its data model, logical file structure, and strategies for document classification and categorization. We also illustrate various retrieval methods and query processing techniques through examples. Finally applications of TEXPROS are presented, where we suggest ways in which the use of the system may alter the software process model.

The authors propose a simple table-free routing technique for the IEEE 802.6 Distributed Queue Dual Bus (DQDB). It offers the simplicity of high-speed packet switching, such as Batcher-Bayan, techniques to metropolitan-area networks. As in high-speed packet switching networks, it consists of a number of stages. A so-called self-routing technique routes a packet from one end to the other without table processing. By checking only one bit in the routing tag of a message at each stage, a switch knows to which switch the message is to be routed in the next stage. The proposed technique is similar to the knockout switch. However, instead of performing switching in a small area, it covers a much wider (metropolitan) area. Therefore, the technique becomes distributed. The proposed routing technique can be extended to cover a service of point-to-multipoint communications, something which is essential for B-ISDN, while maintaining much of its simplicity. Various issues of sequential delivery, replication, and label management are addressed. A novel technique for simplifying topologies is presented.>

Even though IEEE 802.6 can cover longer distances than those of local area networks (LANs), it is limited by unfairness in terms of average access delay. The effect of distance on average access delay is analyzed based on homogeneous assumptions, and a technique to extend 802.6 over a wider area is proposed. The technique makes possible simplification of topology for wide area networks (WANs), which in turn simplifies routing and network management tasks.>

An algorithm for synthesizing a better relational database scheme in elementary key normal form (EKNF) is developed. This algorithm eliminates not only extraneous attributes and other redundancies, but also superfluities from a given set of functional dependences (FDs), based primarily on subset closures, Hamiltonian cycles of FDs, and equivalent subsets of attributes. Following this algorithm, a better LR-minimum FD covering is obtained. A more practical and efficient method for designing a relational database scheme in EKNF is then provided. The time complexity of the algorithm is polynomial.>

THE concepts of system reliability–generally defined as the ability of a system to meet its interface specifications-and of system availability–generally defined as the ability of a system to meet its interface specifications within a specified time limit–predate not only that of distributed computing but also that of the electronic computer itself. With the advent of the electronic computer and its ever increasing penetration of technological, social, and political developments, reliability and availability gained additional recognition as disciplines of serious intellectual challenge. Much of the development of new reliability techniques can be linked directly to the computer hardware developments of the last three decades. However, until relatively recently, the concept of software reliability did not receive significant attention and, when it did, it was rather narrowly focused on approaches to prevent failures; i.e., on software development and verification methodologies, and on languages and tools.

In this paper, we study under what conditions will a pairwise inconsistent relational database ≪R,r≫ have a universal/representative instance L. If R is γ-acyclic and r satisfies all existence constraints, then it is possible to construct a universal instance L, using unmarked nulls, whose total projections onto R yield exactly the relations in r. We show that L would actually be a representative instance under a set of functional dependencies if R satisfies the additional mild condition: for any functional dependency X → A where A is a single attribute, whenever XA is contained in two relation schemes R and R' of R, it follows that R ∩R' is a relation scheme of R, having X as one of its keys.

This paper presents a recognition procedure for parallel tasks in the user program written in a conventional programming language. To establish our program model, it describes the parallelism of the program in tenns of a process flow graph in which the relationships among processes are of predecessors and successors. And finally it presents a parallel processing scheme which realizes automatically the recognition of parallel tasks and schedules these tasks for parallel execution.

We investigate the question of when two entity-relationship diagrams (ERD's) should be considered equivalent, in the sense of representing the same information. This question is very important for a database design process which uses the ERD model, and can be interpreted in various ways. We give three natural and increasingly stricter criteria for developing concepts of equivalence for ERD's. We first give a notion of "domain data compatibility" which ensures that the ERD's in question represent the same universe of data in an aggregate sense. Then we define the set of functional dependencies which are naturally embedded in each ERD, and use it to develop a concept of "data dependency equivalence" which ensures that the ERD's satisfy the same constraints (functional dependencies) among the represented data. Finally, we give our strongest criterion, instance data equivalence, which requires the ERD's to have the same power to represent instances of data. We develop several alternate forms of this third notion, including some giving efficient tableaux tests for its occurrence. Indeed, for each type of equivalence, we give a polynomial-time algorithm to test for it.

The nondeterministic or deterministic entity-relationship model of a database is formaRy defined as a user's view of that database in terms of a collection of time-varying relations: the regular or weak entity relations, or the regular or weak relationship relations. Both nondeterministic and deterministic entity-relationship models have the same strength to characterize information concerning entities and relationships which exist in our minds. An improved table form of the relations is introduced to provide a clear and concise user's view of databases. The basic concept of the entity-relationship approach to the logical database design is provided, and is used to derive 3NF relations. FinaUly, a method of representing physicaly these relations, which are generated by the use of the entity-relationship approach to the logical database design, is presented. Thus, the entity-relationship approach to the logical and physical database design can also be realized.