Source Code Retrieval using Case Based Reasoning

Formal verification of source code has been extensively used in the past few years in order to create dependable software systems. However, although formal languages like Spec# or JML are getting more and more popular, the set of verified implementations is very small and only growing slowly. Our work aims to automate some of the steps involved in writing specifications and their implementations, by reusing existing verified programs. That is, for a given implementation we seek to retrieve similar verified code and then reapply the missing specification that accompanies that code. In this thesis, I present the retrieval system that is part of the Arís (Analogical Reasoning for reuse of Implementation & Specification) project. The overall methodology of the Arís project is very similar to Case-Based Reasoning (CBR) and its parent discipline of Analogical Reasoning (AR), centered on the activities of solution retrieval and reuse. CBR’s retrieval phase is achieved using semantic and structural characteristics of source code. API calls are used as semantic anchors and characteristics of conceptual graphs are used to express the structure of implementations. Finally, we transfer the knowledge (i.e. formal specification) between the input implementation and the retrieved code artefacts to produce a specification for a given implementation. The evaluation results are promising and our experiments show that the proposed approach has real potential in generating formal specifications using past solutions.

Level dc:type.qualificationlevel

masters

Grantor dc:publisher.institution

National University of Ireland Maynooth

Year dc:date.issued

2013

Author dc:creator

• Pitu, Mihai

• Computer Science

Language dc:language

en