Database management

Although the Relational Algebra (RA) is a fundamental component of the relational model of database storage and retrieval, it has not been widely implemented in commercial database systems. The Structured Query Language (SQL) has been for some time the most widely implemented relational database language. This thesis provides and describes the first implementation of a particularly powerful version of the relational algebra that includes an n-ary join operator as well as including the division operator, and nested expressions. With this implementation, that powerful version of the relational algebra can be used as a front end to any SQL-92 compliant database management system, such as Oracle, Informix, DB2, Sybase, SQL Server, and Access.

We critically investigate, evaluate, and illustrate the language of SQL and Java provided by SQLJ. We investigate the main constructs of SQLJ, such as use of declarative and executable statements, SQLJ clauses, and use of bind variables in SQLJ applications. We also observe usage of named and position iterators and we illustrate how to develop and run applications on the server side. In particular, we analyze the difference between SQLJ and JDBC and we find that SQLJ is good for static SQL whereas JDBC is good for dynamic SQL. We also compare the relationship of SQLJ to ANSI/ISO embedded SQL. Finally we evaluate the advantages of developing SQLJ stored procedure over PL/SQL procedures.

This thesis deals with remote database access over the Internet. Three different approaches to remote database access, the CGI-based approach, the Oracle Webserver-based approach and the CORBA-based approach, are discussed. Experiments are conducted to compare the pros and cons of each approach. The Oracle database and Microsoft Access database are utilized as the target database systems. For the CGI-based approach, the CGI programs are coded with ProC, Oracle's SQL-embedded C language, in Microsoft Visual C++ developer's studio, and a Student Record example is demonstrated. For the Oracle Webserver-based approach, the Oracle Webserver 1.1 has been setup and an Online Shopping Club example is demonstrated. For the CORBA-based approach, Orbix Web 2.0.1 and JDK 1.1.1 are utilized as the CORBA system and the development language, respectively, and several examples including single server, multiple servers and multiple databases are presented. An extended SQL syntax is proposed to manipulate data from multiple database systems residing on different hosts in one SQL statement. The implementation of extended SQL statements in CORBA's server/client application is also suggested. With the help of extended SQL syntax and CORBA's multiple server access feature, an example of the client application accessing two databases at a time is illustrated. The pros and cons are discussed of the three remote database access approaches studied in the thesis: (I) The CGI-based approach, which is the most generic and widely used approach. (II) The Oracle Webserver approach, which is a relatively simplified way of the CGI approach, easy to apply but limited in applications. (III) The CORBA-based approach, which is the most powerful and sophisticated approach for remote database access.

We survey and critically compare various approaches and software products for the modeling and logical design of relational databases. In particular, we examine different information modeling methodologies, such as Information Engineering, CASE$\sp*$Method, the Three Schema Concept, as well as data modeling techniques, such as the Entity-Relationship techniques, and the Object Role Modeling technique.

This thesis, based on (ISO/ANSI) X3H2-93-091 Working Draft Database Language SQL (SQL3) (Melton, 1993, 1994), systematically specifies the characteristics and manipulation of object-oriented techniques in SQL3. An essential difference between SQL3 and the previous SQL standards lies in the fact that the abstract data type (ADT) facility supported in SQL3 causes SQL3 to become an extended relational data model rather than a "pure" relational data model. This master's thesis focuses on the careful examination and explication of ADTs and the corresponding families--type templates as they are used in SQL3. These topics involve four aspects: (1) general concepts and technologies in object-oriented data models, (2) the definition of ADTs and their advantages, (3) the manipulation of ADTs in SQL3, including type templates (e.g., distinct types, explicit ADTs, user-defined type templates, standard type templates) and collection types, (4) "tables of ADTs" (row types with table), and (5) conclusions and challenges.

One of the major components of any pervasive system is its proactive behavior. Various models have been developed to provide system wide changes which would enable proactive behavior. A major drawback of these approaches is that they do not address the need to make use of existing applications whose design cannot be changed. To overcome this drawback, a middleware architecture called "Concord" is proposed. Concord is based on a simple model which consists of Lookup Server and Database. The rewards for this simple model are many. First, Concord uses the existing computing infrastructure. Second, Concord standardizes the interfaces for all services and platforms. Third new services can be added dynamically without any need for reconfiguration. Finally, Concord consists of Database that can maintain and publish the active set of available resources. Thus Concord provides a solid system for integration of various entities to provide seamless connectivity and enable proactive behavior.

This research focuses on analyzing the effect of change in network parameters on the performance of GSM cellular network. A simulation model is created using visual C++ and network performance in terms of call failures and handover failures are studied. A cellular network architecture and its performance are based on so many parameters and complexities that it is very difficult to analyze the effect of a parameter on any one aspect of performance. The purpose of this thesis is to study the performance of network with respect to some of those parameters while neglecting other effects. The simulated network takes parameters like cell diameter, traffic intensity and mobility pattern as input and a simulation run results in the evaluation of the call failure and success statistics.

Collaborative filtering (CF), a very successful recommender system, is one of the applications of data mining for incomplete data. The main objective of CF is to make accurate recommendations from highly sparse user rating data. My contributions to this research topic include proposing the frameworks of imputation-boosted collaborative filtering (IBCF) and imputed neighborhood based collaborative filtering (INCF). We also proposed a model-based CF technique, TAN-ELR CF, and two hybrid CF algorithms, sequential mixture CF and joint mixture CF. Empirical results show that our proposed CF algorithms have very good predictive performances. In the investigation of applying imputation techniques in mining incomplete data, we proposed imputation-helped classifiers, and VCI predictors (voting on classifications from imputed learning sets), both of which resulted in significant improvement in classification performance for incomplete data over conventional machine learned classifiers, including kNN, neural network, one rule, decision table, SVM, logistic regression, decision tree (C4.5), random forest, and decision list (PART), and the well known Bagging predictors. The main imputation techniques involved in these algorithms include EM (expectation maximization) and BMI (Bayesian multiple imputation).