Kaushal Chari

Software Effort, Quality, and Cycle Time: A Study of CMM Level 5 Projects

The Capability Maturity Model (CMM) has become a popular methodology for improving software development processes with the goal of developing high-quality software within budget and planned cycle time. Prior research literature, while not exclusively focusing on CMM level 5 projects, has identified a host of factors as determinants of software development effort, quality, and cycle time. In this study, we focus exclusively on CMM level 5 projects from multiple organizations to study the impacts of highly mature processes on effort, quality, and cycle time. Using a linear regression model based on data collected from 37 CMM level 5 projects of four organizations, we find that high levels of process maturity, as indicated by CMM level 5 rating, reduce the effects of most factors that were previously believed to impact software development effort, quality, and cycle time. The only factor found to be significant in determining effort, cycle time, and quality was software size. On the average, the developed models predicted effort and cycle time around 12 percent and defects to about 49 percent of the actuals, across organizations. Overall, the results in this paper indicate that some of the biggest rewards from high levels of process maturity come from the reduction in variance of software development outcomes that were caused by factors other than software size

Demystifying integration

This framework helps navigate the standards maze to develop a flexible applications infrastructure that advances an organizations strategic business needs.

Model composition in a distributed environment

Organizations that operate from multiple locations have data and model resources that are distributed at various sites of the organization. Decision making is facilitated when these resources are leveraged to support model composition and execution, i.e., composing and executing a sequence of models in response to a particular decision making situation. In this paper, we address the problem of model composition when data sources (models and data) are distributed across multiple sites and have different scopes.

Information Market-Based Decision Fusion

Improved classification performance has practical real-world benefits ranging from improved effectiveness in detecting diseases to increased efficiency in identifying firms that are committing financial fraud. Multiclassifier combination (MCC) aims to improve classification performance by combining the decisions of multiple individual classifiers. In this paper, we present information market-based fusion (IMF), a novel multiclassifier combiner method for decision fusion that is based on information markets. In IMF, the individual classifiers are implemented as participants in an information market where they place bets on different object classes. The reciprocals of the market odds that minimize the difference between the total betting amount and the potential payouts for different classes represent the MCC probability estimates of each class being the true object class. By using a market-based approach, IMF can adjust to changes in base-classifier performance without requiring offline training data or a static ensemble composition. Experimental results show that when the true classes of objects are only revealed for objects classified as positive, for low positive ratios, IMF outperforms three benchmarks combiner methods, majority, average, and weighted average; for high positive ratios, IMF outperforms majority and performs on par with average and weighted average. When the true classes of all objects are revealed, IMF outperforms weighted average and majority and marginally outperforms average.

Model Composition Using Filter Spaces

Decision support systems (DSS) typically contain data and models to facilitate decision making. DSS users, in response to a particular decision-making situation, often execute a sequence of models, in which inputs to a model in the sequence are obtained from outputs of other models upstream in the sequence and from database retrievals. The problem of generating a sequence of models from the set of available models is known as the model composition problem. In this paper, we propose a new construct called filter spaces to support model composition. We show how filter spaces can significantly facilitate automation of model composition and execution process, and provide effective means to integrate partial solutions from multiple composite models and databases.

Negotiation Behaviors in Agent-Based Negotiation Support Systems

Prior research on negotiation support systems (NSS) has paid limited attention to the information content in the observed bid sequences of negotiators as well as on the cognitive limitations of individual negotiators and their impacts on negotiation performance. In this paper, we assess the performance of human subjects in the context of agent-based NSS, and the accuracy of an exponential functional form in representing observed human bid sequences. We then predict the reservation values of negotiators based on their observed bids. Finally, we study the impact of negotiation support systems in helping users realize superior negotiation outcomes. Results indicate that an exponential function is a good model for observed bids.

Resource allocation and capacity assignment in distributed systems

This paper considers the problem of assigning computers, database files and communication facilities in Distributed Computing Systems (DCS). A comprehensive mathematical model is presented that incorporates the trade-offs involved in the design of computer networks and database files. An efficient heuristic procedure is developed that considers simultaneously the allocation of computers, database files, programs for generating query reports, and communication facilities. Heuristic solutions for a variety of test problems are presented and compared with various benchmarks.

Multi-Issue Automated Negotiations Using Agents

Software agents can perform effectively as negotiators in automated negotiation settings. We present a model for software agents that can automate negotiations by implementing a multi-issue learning heuristic that allows agents to learn from the bidding behavior of opponents. The performance of agents is evaluated using an experimental study involving human subjects. The results indicate that software agents can act as effective surrogates of human negotiators under some circumstances.

An implementation of a graph-based modeling system for structured modeling (GBMS/SM)

Abstract This paper describes GBMS/SM, a graph-based modeling system based on structured modeling that supports model formulation, maintenance and solution. Key features of GBMS/SM include the following: (1) model construction using graphical inputs, (2) modeling in multiple problem domains, (3) syntax-directed editing, (4) model views at different levels of detail, (5) automation of model data acquisition, and (6) interfaces to solvers, spreadsheets and databases. This paper presents design and implementation details of GBMS/SM that are of general interest to graphical modeling system designers. Challenges that were encountered in the development of GBMS/SM are emphasized in this paper.

System Test Planning of Software: An Optimization Approach

This paper extends an exponential reliability growth model to determine the optimal number of test cases to be executed for various use case scenarios during the system testing of software. An example demonstrates a practical application of the optimization model for system test planning