An improved self organizing map using jaccard new measure for textual bugs data clustering
In software projects there is a data repository which contains the bug reports. These bugs are required to carefully analyze to resolve the problem. Handling these bugs humanly is extremely time consuming process, and it can result the delaying in addressing some important bugs resolutions. To overc...
محفوظ في:
| المؤلف الرئيسي: | |
|---|---|
| التنسيق: | أطروحة |
| اللغة: | English English English |
| منشور في: |
2018
|
| الموضوعات: | |
| الوصول للمادة أونلاين: | http://eprints.uthm.edu.my/517/1/24p%20ATTIKA%20AHMED.pdf http://eprints.uthm.edu.my/517/2/ATTIKA%20AHMED%20COPYRIGHT%20DECLARATION.pdf http://eprints.uthm.edu.my/517/3/ATTIKA%20AHMED%20WATERMARK.pdf |
| الوسوم: |
إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
|
| الملخص: | In software projects there is a data repository which contains the bug reports. These bugs are required to carefully analyze to resolve the problem. Handling these bugs humanly is extremely time consuming process, and it can result the delaying in addressing some important bugs resolutions. To overcome this problem researchers have been introduced many techniques. One of the techniques is the bug clustering. For the purpose of clustering, a variety of clustering algorithms available. One of the commonly used algorithm for bug clustering is K-means, which is considered a simplest unsupervised learning algorithm for clustering, yet it tends to produce smaller number of cluster. Considering the unsupervised learning algorithms, Self-Organizing Map (SOM) considers the equally compatible algorithm for clustering, as both algorithms are closely related but different in way they were used in data mining. This research attempts a comparative analysis of both the clustering algorithms and for attaining the results, a series of experiment has been conducted using Mozilla bugs data set. To address the data sparseness issue, the experiment has been performed on textual bugs’ data by using two different distance measure which are Euclidean distance and Jaccard New Measure. The research results suggested that SOM has a limitation of poor performance on sparse data set. Thus, the research introduced the improved SOM algorithm by using a Jaccard NM (SOM-JNM). The SOM-JNM produced significantly better results therefore; it can be consider a challenging approach to address the sparse data problems. |
|---|