Luigi Benedicenti

Analysis of the effects of software reuse on customer satisfaction in an RPG environment

This paper reports on empirical research based on two software products. The research goal is to ascertain the impact of the adoption of a reuse policy on customer satisfaction. The results show that when a systematic reuse policy is implemented, such as the adoption of a domain specific library: reuse is significantly positively correlated with customer satisfaction; and there is a significant increase in customer satisfaction. The results have been extended to the underlying populations, supposed normal.

BAYESIAN NETWORK BASED XP PROCESS MODELLING

A Bayesian Network based mathematical model has been used for modelling Extreme Programming software development process. The model is capable of predicting the expected finish time and the expected defect rate for each XP release. Therefore, it can be used to determine the success/failure of any XP Project. The model takes into account the effect of three XP practices, namely: Pair Programming, Test Driven Development and Onsite Customer practices. The model’s predictions were validated against two case studies. Results show the precision of our model especially in predicting the project finish time.

Operational Wellness Model: A Wellness Model Designed for an Agent-Based Wellness Visualization System

This paper introduces an operational wellness model that forms the basis of a wellness visualization system. The operational wellness model is needed to determine a common reference enabling the evaluation of the visualization system. Existing definitions of wellness and wellness evaluation models have limitations and cannot be realized in software. The proposed model comprises the operational definition of wellness and the operational wellness evaluation model. The results generated by the operational wellness model will be used in the wellness visualization system. The model presents numerous advantages: it is extensible, computable, and can be rendered by a computer program.

Agent behavior and agent models in unregulated markets

Mobile-agent systems show significant promise as the most effective way to harness the power of the Internet and the massive collection of information and opportunity that the Internet holds. However the efficient organization and control of these systems remains one of a number of unsolved problems with this approach to network computing. This paper examines a mobile-agent system with specific focus on environment sensing, preemptive load balancing and open agent markets. Agent behaviour is studied with actual agent systems using progressively sophisticated agent migration strategies.It is shown that actual modeling shows interesting and difficult to predict behaviour in the agent systems. It is shown that mobile agents with relatively simple migration strategies can cause loads in self-regulating agent markets to oscillate. It is further shown that using Autoregressive modeling to predict the market behaviour can allow individual agents to significantly outperform other agents. However the fidelity of the model is critical to the success of the agents. The criticality of good agent strategies and actual agent system modeling is thus highlighted.

An examination of mobile agents system evolution in the course scheduling problem

This paper examines the evolution of a mobile agent system as a mechanism for solving the course-scheduling problem. In our agent model, each mobile agent represents a course (named course agent). Course agents negotiate with each other through a mechanism we define as a signboard agent. There is a signboard agent for each day of the week. Each platform represents a weekday. In the paper, the mobile agents technology is based on the TEEMA (TRLabs Execution Environment for Mobile Agents) platform. We believe that this agent approach to solving the course-scheduling problem is unique. Our initial results indicate that this approach can lead to an effective and efficient solution to the problem.

Establishing quality control in software agents

This paper illustrates the results of a research effort to measure the quality of software agents written in Java using extreme programming. A factoring of quality has been chosen in accordance with the properties and limitations of the type of products and development method considered. The use of the representational theory of measurement gave quantitative measures. The measure itself is then used to identify quality clusters that define an ordinal quality taxonomy. The result is a measure of quality that depends only on code, that is easy to implement and to take, and that also gives an indication of the improvements to low-quality classes.

Object oriented process modeling with fuzzy logic

Fuzzy Logic has traditionally found an application in control theory, systems analysis, and artificial intelligence [9] [7]. However, it is very difficult to apply it to traditional software engineering. There is evidence that fuzzy logic can help in avoiding early assignment errors in SW engineering methods [1]. The authors have developed a technique that allows object oriented business process modeling. In this paper they propose a fuzzy logic extension to the method. Fuzzy sets can depict the uncertainty on the cost driver. Thus cost driver uncertainties have less impact on the model. The approach is new since it applies fuzzy logic to process modeling, and not to control or expert systems.

Improved Agile: A Customized Scrum Process for Project Management in Defense and Security

This chapter describes our experience of adopting agile project management in a software development project in the defense context. Adopting an agile approach for both software development and the management of the consequent project is becoming increasingly more relevant in many domains, and the defense domain is no exception. However, this relevance does not by itself facilitate adoption as the defense environment is sometimes characterized as an unyielding culture where change is very difficult to effect. We addressed this characterization by obtaining strong unequivocal support from the top leadership, extending the standard scrum roles to fit within a hierarchical organization, and creating mixed teams of civilian consultants and military officers. This strategy was further supported by dedicated infrastructure consisting of specialized training, a suite of computer-aided software engineering tools, a structured governance community, and a custom agile doctrine. As a result of the introduction of our method, we obtained a marked improvement in development costs and project scope with a consequent sharp increase of customer satisfaction. The product has already been tested in an inter-force simulation with excellent results, which further confirms the feasibility of our method for mission-critical software systems for the Italian Armed Forces.

An agent based general solution model for the course timetabling problem

This paper proposes a general solution model using agent technology in order to solve two critical open issues in the course timetabling problem: application-specific issues and dynamic-constraint issues. In our solution model, each agent in the system represents a constraint in the course timetabling problem. Compared to existing course timetabling solutions, the proposed agent solution model has much better generality, flexibility, dynamics, modularity and scalability.

Mobile agent network management system performance study in frame relay network

This paper illustrates the results of a research effort to apply mobile agent technology in network management for the frame relay network The paper implemented a Java based mobile agent network management approach in the frame relay network by using Teema, a unique agent execution platform. The agent network management performance is simulated in a number of scenarios. It is then contrasted with the SNMP client-server paradigm. The bandwidth expenditure and cost in each scenario is compared and contrasted to the client-server paradigm. The results in the paper suggest that mobile agent technology is well suited to address some client-server management paradigm limitations on frame relay network topology and that the agents have better performance in a series of cases.