W. Homer Carlisle

A self-improving helpdesk service system using case-based reasoning techniques

Abstract Case-Based Reasoning (CBR) is the process of solving a given problem based on the knowledge gained from solving precedents. It is an effective technique in the area of customer services or helpdesks. That is, a CBR system is used to solve most of the commonly occurring customer problems. While the implementation techniques may vary, most CBR systems include the following five steps: case representation and storage, precedent matching and retrieval, adaptation of the retrieved solution, validation of the solution, and finally, casebase update to include the information gained from the new problem. This paper details the various implementation techniques for these five steps, while focusing on a particular helpdesk system, namely SmartUSA, developed for the Union Camp Corporation. This system solves a customers problem by filtering the problem description through an alias table to generate a brief description and then matching the brief description with the cases in the database. It has proved to be an effective and user-friendly system that has successfully handled different descriptions of the same problem and allowed for the casebase to be built in free-format (plain) text. This system has significantly reduced the workload and the response time in the customer services department of the Union Camp Corporation.

A PERFORMANCE EVALUATION OF HEURISTICS-BASED TEST CASE GENERATION METHODS FOR SOFTWARE BRANCH COVERAGE

Software testing is an important step in the development of complex systems. The construction of test cases using traditional methods usually requires considerable manual effort. QUEST/Ada—Query Utility Environment for Software Testing of Ada, is a prototype test case generation system that uses various heuristics-based approaches to generate test cases. The system, which is designed for unit testing, generates test cases by monitoring the branch coverage progress and intelligently modifying existing test cases to achieve additional coverage. Three heuristics-based approaches along with a random test case generation method were studied to compare their branch coverage performance. Although some constraints are imposed by the prototype, the framework provides a useful foundation for further heuristics-based test case generation research. The design of the system, the heuristic rules used in the system, and an evaluation of each rule’s performance are presented.

A heuristic approach for test case generation

Test case generation using traditional software testing methods generally requires considerable manual effort and generates only a limited number of test cases before the amount of time expended becomes unacceptably large. A rule-based framework that will automatically generate test data to achieve maximal branch coverage is presented. The rationale of the heuristic rules and the strategy for the test case generation are also described. The result of this approach shows its potential for improving software testing. The rule-based approach allows this framework to be extended to include additional testing requirements and test case generation knowledge.

A framework for intelligent test data generation

Test data generation using traditional software testing methods generally requires considerable manual effort and generates only a limited number of test cases before the amount of time expanded becomes unacceptably large. A rule-based framework that will automatically generate test data to achieve maximal branch coverage is presented. The design and discovery of rules used to generate meaningful test cases are also described. The rule-based approach allows this framework to be extended to include additional testing requirements and test case generation knowledge.

Expert system assisted test data generation for software branch coverage

Abstract With the increased production of complex software systems, verification and validation (V & V) has evolved into a set of activities that span the entire software life cycle. Among these various activities, software testing plays a major role in V&V. Conventional software testing methods generally require considerable manual effort which can generate only a limited number of test cases before the amount of time expended becomes unacceptably large. In this paper, we present a new approach to generating test cases based on artificial intelligence methods. By analyzing the branch coverage of previous test cases, an expert system is able to generate new test cases which provide additional coverage. Heuristic rules are used to modify previous test cases in order to achieve the desired branch coverage. This approach to software testing has the potential for greatly reducing the overall costs associated with branch coverage testing.