Automated recognition of algorithms from source code

No Thumbnail Available
Journal Title
Journal ISSN
Volume Title
Helsinki University of Technology | Diplomityö
Checking the digitized thesis and permission for publishing
Instructions for the author
Date
2009
Major/Subject
Ohjelmistotekniikka
Mcode
T-106
Degree programme
Language
en
Pages
55
Series
Abstract
Automated recognition of algorithms from source code plays more and more important role in software maintenanee, programming education and visualization programming. The thesis first defines that automated recognition of algorithms is a process of discovering the design or the purpose behind a complex source code or even a system, by analyzing, abstracting information from source code and deductively applying ready known plan templates. Then, it reviews five classical cognition models and three previous prototypes based on the cognition models. The thesis also introduces the design and algorithms of an automated program recognizer named SACR, which implemented plan based abstract concepts recognition. By presenting and analyzing the experimental data, the thesis shows that the plan based recognition of the abstract concepts is effective and accurate enough to recognize algorithms from source code automatically. Finally, the thesis proposes several potential improvements for the future work.
Description
Supervisor
Malmi, Lauri
Keywords
program comprehension, automated program comprehension, program understanding, cognition models, abstract concepts, graph parsing, roles of variables, SACR, plan, AST, similarity
Other note
Citation