This is a Demo Server. Data inside this system is only for test purpose.

Browsing by Author "Ufuktepe, E."

Filter results by typing the first few letters
Now showing 1 - 3 of 3
  • Thumbnail Image
    Conference Object
    Citation Count: 4
    Code change sniffer: Predicting future code changes with Markov chain
    (Institute of Electrical and Electronics Engineers Inc., 2021) Ufuktepe, E.; Tuglular, T.; Tuğlular, Tuğkan; Bilgisayar Mühendisliği Bölümü
    Code changes are one of the essential processes of software evolution. These changes are performed to fix bugs, improve quality of software, and provide a better user experience. However, such changes made in code could lead to ripple effects that can cause unwanted behavior. To prevent such issues occurring after code changes, code change prediction, change impact analysis techniques are used. The proposed approach uses static call information, forward slicing, and method change information to build a Markov chain, which provides a prediction for code changes in the near future commits. For static call information, we utilized and compared call graph and effect graph. We performed an evaluation on five open-source projects from GitHub that varies between 5K-26K lines of code. To measure the effectiveness of our proposed approach, recall, precision, and f-measure metrics have been used on five open-source projects. The results show that the Markov chain that is based on call graph can have higher precision compared to effect graph. On the other hand, for small number of cases higher recall values are obtained with effect graph compared to call graph. With a Markov chain model based on call graph and effect graph, we can achieve recall values between 98%-100%. © 2021 IEEE.
  • Thumbnail Image
    Conference Object
    JavaScript kütüphaneleri için girdi dogrulama analizi
    (CEUR-WS, 2015) Ufuktepe, E.; Tuglular, T.; Tuğlular, Tuğkan; Bilgisayar Mühendisliği Bölümü
    Nowadays, mobile and web based software has been an integral part of our lives. In recent years, there has been an increase in usage of JavaScript libraries in those kind of software. Although these JavaScript libraries fulfill the functions they have promised with respect to the application program interfaces they provide, they are not robust against unexpected inputs. In this study, the parameters and global variables of functions in the five selected JavaScript li-braries that are frequently used in mobile and web based software are analyzed for input validation. For this purpose, an input validation model has been pro-posed. Based on this model, a tool called TAJS that performs a type analysis on JavaScript programs has been extended with a proposed algorithm. The result-ing tool is executed on five JavaScript libraries and obtained results are shared.
  • Thumbnail Image
    Conference Object
    Citation Count: 8
    A program slicing-based Bayesian network model for change impact analysis
    (Institute of Electrical and Electronics Engineers Inc., 2018) Ufuktepe, E.; Tuglular, T.; Tuğlular, Tuğkan; Bilgisayar Mühendisliği Bölümü
    Change impact analysis plays an important role in identifying potential affected areas that are caused by changes that are made in a software. Most of the existing change impact analysis techniques are based on architectural design and change history. However, source code-based change impact analysis studies are very few and they have shown higher precision in their results. In this study, a static method-granularity level change impact analysis, that uses program slicing and Bayesian Network technique has been proposed. The technique proposes a directed graph model that also represents the call dependencies between methods. In this study, an open source Java project with 8999 to 9445 lines of code and from 505 to 528 methods have been analyzed through 32 commits it went. Recall and f-measure metrics have been used for evaluation of the precision of the proposed method, where each software commit has been analyzed separately. © 2018 IEEE.