CODECOD: Platform Crowdsourcing Untuk Mendeteksi Code Smell

• Main Author: JAMIATI PARAMITA NIM - 23516023 , ANDI
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/25499
• Institution: Institut Teknologi Bandung

One phase of software development is maintenance. The ease of maintenance is supported by the program code written in the right way. However, sometimes programmers still have bad habits in writing code, causing code smell. For that the detection of the code smell in the program code must be done as early as possible with the correct accuracy. At this time many code-detection tools smell in the program code, but the role of humans is still very much needed to do manual detection. One method that can be done to combine human roles and machine computing is the crowdsourcing method. However, research on crowdsourcing in the field of software engineering in general and the detection of smell code in particular is still very minimal. <br /> <br /> <br /> <br /> <br /> From these problems, this research develops one of the crowdsourcing methods, namely the decomposition of tasks implemented in a platform called CODECOD. This platform helps humans to detect smell code according to the classification of smell code from Martin Fowler by decomposing the program code to help detect. In addition, this study also developed the find, vote method, verify as a quality guarantee for the detection of code smell. <br /> <br /> <br /> <br /> <br /> The implementation of this research is decomposing a java program code file with object-oriented structure into a single method and single class which is then referred to as microtask. Based on the results of the test, the detection of the smell code using CODECOD shows that the time required is less, the amount of code smell found is more and the accuracy is greater than without decomposition. <br />