Pai-Cheng Chu

Statistical profile estimation in database systems

A statistical profile summarizes the instances of a database. It describes aspects such as the number of tuples, the number of values, the distribution of values, the correlation between value sets, and the distribution of tuples among secondary storage units. Estimation of database profiles is critical in the problems of query optimization, physical database design, and database performance prediction. This paper describes a model of a database of profile, relates this model to estimating the cost of database operations, and surveys methods of estimating profiles. The operators and objects in the model include build profile, estimate profile, and update profile. The estimate operator is classified by the relational algebra operator (select, project, join), the property to be estimated (cardinality, distribution of values, and other parameters), and the underlying method (parametric, nonparametric, and ad-hoc). The accuracy, overhead, and assumptions of methods are discussed in detail. Relevant research in both the database and the statistics disciplines is incorporated in the detailed discussion.

Validating the Computer Anxiety Rating Scale: Effects of Cognitive Style and Computer Courses on Computer Anxiety.

Abstract In this paper, the validity of a newly developed computer anxiety questionnaire, the Computer Anxiety Rating Scale (CARS) of Heinssen, Glass, and Knight (1987), is tested. Expectations are developed about how a valid measure of computer anxiety would behave in response to several demographic traits and cognitive styles, and evidence is gathered about the expectations. Factor analysis is used to identify factors of computer anxiety underlying the CARS. In general, the CARS is found to be a valid instrument. Four items in the CARS, however, systematically behave differently from expectations and therefore may not properly measure computer anxiety. Cognitive style is shown to be an important determinant of computer anxiety, with intuitive and thinking individuals exhibiting lower anxiety than their sensing and feeling counterparts. Computer courses may be useful in reducing anxiety for some anxiety factors for some cognitive styles.

The Joint Effects of Effort and Quality on Decision Strategy Choice with Computerized Decision Aids

Research has recently focused on the effort-reduction or minimization role of computerized decision aids, and how users may employ aids to manage their effort, which in turn affects their choice of decision strategies. In this paper, it is argued that consideration of effort reduction or minimization by itself is not sufficient for inducing changes in decision strategy. Instead, decision aid effects on effort must be considered jointly with the decision quality associated with the various decision strategies. This is true even if the decision aid has no effect on decision quality. We adapt and extend a theoretical framework that can be used to evaluate the joint effects of effort and quality on decision strategy choice. In addition, we reinterpret past research results in light of the framework and present new experimental evidence on the descriptive validity of the framework.

Perceptions of accuracy and effort of decision strategies

Abstract A critical tenet of cost-benefit theories of decision strategy choice is that decision makers’ perceptions of accuracy and effort determine strategy selection. However, little research has been conducted on human perceptions of decision strategy accuracy and effort. Instead, researchers have substituted deductively derived inferences on strategy accuracy and effort for perceptions in interpreting decision processes. In this study, we used a survey to study perceptions of decision strategies. The results indicate that participants as a group understood the accuracy and effort dimensions of decision strategies. The participants’ perceptions of the accuracy and effort of various decision strategies largely agreed with researchers’ deductions. However, there was substantial variation across individuals in perceptions of various decision strategies and in the composition of efficient frontiers of decision strategies.

A transaction-oriented approach to attribute partitioning

Abstract This study introduces the concepts of sufficient and support to design two methods for solving the attribute partitioning problem: MAX and FORWARD SELECTION. These concepts express the simple idea that in order to minimize costs, a segment should contain all the attributes required by some transaction(s) and exclude other attributes. Contrary to all previous studies which without exception treat the attribute as the decision variable and may become computationally infeasible when the number of attributes is large (a condition favoring attribute partitioning), these two methods treat the transaction as the decision variable and their run time is not affected by the number of attributes. Both methods generate excellent results without using complicated mathematics. They also take into account the complexity caused by the interaction between attribute partitioning and access path selection, an issue most previous studies fail to address. Besides, both methods provide detailed information normally not available from other methods. This information provides new insight into the process of attribute partitioning. MAX is for cases involving a modest number of transactions (⩽15), FORWARD SELECTION for cases involving a large number of transactions.

Induced system restrictiveness: an experimental demonstration

Decision support systems (DSSs) built to expand human decision capabilities can simultaneously restrain decision-making. This characteristic is referred to as system restrictiveness. The existing conceptualization of system restrictiveness is reviewed, and the concept of induced restrictiveness as a subtle force that restricts decision-making by inducing decision-makers to take a particular decision process is introduced. The validity of this concept has been demonstrated by a controlled laboratory experiment that employed a protocol analysis technique. The experiment evaluated the decision process induced by Lotus 1-2-3 for a particular type of task. The results indicated that Lotus tended to induce the incremental decision process as opposed to the synoptic decision process favored by the control group. The incremental decision process generates alternatives by making marginal changes to a previous solution, whereas the synoptic decision process is characterized by unbiased search for solutions. These results are discussed in the context of induced restrictiveness. >

Decision Processes and Use of Decision Aids Comparing Two Closely Related Nations in East Asia

The authors report the results of a controlled laboratory experiment comparing the decision processes of participants from Taiwan and Japan. These two nations have very close geographical, cultural, historical, and economic ties. The results show that decision processes of Taiwanese differed from those of Japanese. Specifically, decision processes adopted by Taiwanese participants were more consistent with compensatory processes than were those adopted by Japanese participants. The authors identify cultural factors that may explain the differences. These results demonstrate the danger of generalizing decision theories across national boundaries, even when the nations are seemingly closely related. The results also indicate that the differences in decision processes among nations cannot easily be characterized as East versus West.

Cell suppression methodology: the importance of suppressing marginal totals

Safeguarding confidential information is of paramount concern to government agencies in publishing statistical data. Given a set of sensitive cells, the problem is to identify a set of complementary cells to suppress so as to mask the values of the sensitive cells. All of the existing cell suppression methods fail to consider the relationships among cell values and the representation of these relationships in marginal totals. That marginal totals may contain potent information has not been appreciated. The paper employs the theory of nominal data analysis to demonstrate that the disclosure of marginal totals can be very risky. It recommends adding a front end test to the existing methods. The goal is to identify a list of sensitive marginal totals that have to be suppressed. This increases the sophistication of cell suppression methodology by providing an extra layer of protection.

An object-oriented approach to modeling financial accounting systems

Abstract This paper analyzes the requirements of a financial accounting system and compares an object-oriented data model with the relational data model as a vehicle for meeting these requirements. It is shown that the relational data model, despite its market dominance, is not an effective tool for modeling financial accounting systems. Its deficiency stems from a want of intrinsic capability to model generalizing abstractions and complex structures and a lack of control over accounting procedures. In contrast, an object-oriented model supports generalizing abstractions and complex data structures. Also, the encapsulation of behavior within an object provides a means for modeling accounting procedures. These features promote a design that integrates accounting structures and accounting procedures and help create a system that is friendly, modular, and reliable. This paper also proposes the idea of establishing a library of accounting objects as building blocks to configure new accounting systems. A discussion of some future research directions concludes the paper.

The impact of problem size on decision processes: an experimental investigation on very large choice problems with support of decision support systems

Abstract: Choice problems as a class of decision problems have attracted great attention for the last couple of decades. Among the frameworks and supporting theories used in their study, two have had the greatest impact: bounded rationality and cost–benefit. Both theories could find support from past empirical studies under different conditions or problem environments. In the past studies, problem size has been shown to play an important role in decision-making. As problem size increases, a decision process may be detoured and the decision outcome may be different. In this paper we investigate the impact of problem size on three important aspects of the computer-aided decision process – strategy selection, decision time/effort, and decision quality – through very large choice problems.