Functional program testing and analysis
Functional program testing and analysis
Requirements Engineering: The State of the Practice
IEEE Software
Market research for requirements analysis using linguistic tools
Requirements Engineering
Testing against Natural Language Requirements
QSIC '07 Proceedings of the Seventh International Conference on Quality Software
Text2Test: Automated Inspection of Natural Language Use Cases
ICST '10 Proceedings of the 2010 Third International Conference on Software Testing, Verification and Validation
NLDB'05 Proceedings of the 10th international conference on Natural Language Processing and Information Systems
Hi-index | 0.00 |
Generating Test Cases from natural language requirements pose a formidable challenge as requirements often do not follow a defined structure. In this paper, we present a tool to generate Test Cases from a functional requirement document. No restriction on the structure of the sentence is imposed. The tool works on each requirement sentence and generates one or more Test Cases through a five step process --- 1) The sentence is analyzed through a syntactic parser to identify whether it is testable; 2) A compound or complex testable sentence is split into individual simple sentences; 3) Test Intents are generated from each simple sentence (Test Intents map to the aspects on which the requirement is to be tested); 4) The Test Intents are grouped and sequenced in temporal order to generate Positive Test Cases. A Positive Test Case verifies the affirmative action of the system; 5) Wherever applicable, Boundary Value Analysis and other techniques are used generate Negative Test Cases. Negative Test Cases verifies the behavior of the system in exception conditions. The automated generation of the Test Cases has been implemented in a tool called Litmus. We provide experimental results of our tool on actual requirement documents across domains and discuss the advantages and shortcomings of our approach.