Sarnaste koodisegmentide põhjal soovitusmudeli loomine
Date
2016
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Käesoleva töö eesmärgiks on luua töövoog soovitusmudeli koostamiseks, mida saaks rakendada staatilises koodianalüüsis kasutajale soovituste tegemisteks. Selleks otsib süsteem sarnaseid koodisegmente avatud lähtekoodiga projektidest ning klasterdab need sarnasuse alusel. See lähenemine põhineb hüpoteesil, et kui arendajad kirjutavad mitu korda sarnast koodi, siis sisaldab see programmeerimismustrit, mida võiks teistelegi soovitada. Töös kontrollitakse lähenemise sobivust ning hinnatakse loodud meetodi kasulikkust staatilises koodianalüüsis. Kokkuvõttes leiti, et sel viisil tehtud soovitused ei arvesta piisavalt analüüsitava projekti eripäradega ega pole rakendatavad selles projektis.
The goal of this work is to develop a workflow for the creation of recommendation model which could be used in static code analysis. The system searches for similar code segments from open source projects and clusters them based on their similarity. The hypothesis is that whenever developers write same code multiple times, it would contain an useful pattern that can be recommended to others. The aim of this work is to test this assumption and evaluate the usefulness of the developed process for static code analysis. In conclusion, it was found that the recommendations do not consider the specifics of the analysed project and are therefore not applicable in the project.
The goal of this work is to develop a workflow for the creation of recommendation model which could be used in static code analysis. The system searches for similar code segments from open source projects and clusters them based on their similarity. The hypothesis is that whenever developers write same code multiple times, it would contain an useful pattern that can be recommended to others. The aim of this work is to test this assumption and evaluate the usefulness of the developed process for static code analysis. In conclusion, it was found that the recommendations do not consider the specifics of the analysed project and are therefore not applicable in the project.