Faster Detecting of Duplicate Bug Report with Maintaining Accuracy

Aminoroaya, Zahra; Soleimani Neysiani, Behzad; Nadimi Shahraki, Mohammad Hossein

Faster Detecting of Duplicate Bug Report with Maintaining Accuracy

نویسندگان

Zahra Aminoroaya ¹

Behzad Soleimani Neysiani ²

Mohammad Hossein Nadimi Shahraki ³

¹ Allame Naeini Higher Education Institute, Naein, Isfahan, Iran

² Faculty of Electrical and Computer Engineering, University of Kashan, Kashan, Isfahan, Iran

³ Faculty of Computer Engineering, Najafabad Branch Islamic Azad University, Najafabad, Isfahan, Iran

چکیده

Nowadays, duplicate bug reports detection is one of major problems for user bug reports tracking systems. Many researchers used information retrieval tools and methods for solving this problem which are used in this study with new feature extracted based on minimum, maximum and average similar frequent terms in two bug reports. First,162 features are extracted in 4 large data set repositories as Android, Mozilla, Open Office and Eclipse which
contains of new and introduced features in the state of the art. Then dimension reduction methods are used to eliminate negligible importance features and improve the running time of train and testing classification algorithms.Experimental results show the runtime time of the classification algorithms by reduced features reduced from minutes to seconds in comparing to all features, while the accuracy of duplicate detection improved between 1% to 6%. Due to new proposed features, the accuracy and recall of duplicate detection are more than 96% and 90% respectively.

کلیدواژه‌ها

Feature Extraction

Duplicate Bug Reports Detection

Classification Algorithms

Accuracy

Run-Time