Zarinah Mohd Kasirun

Why so complicated? Simple term filtering and weighting for location-based bug report assignment recommendation

Large software development projects receive many bug reports and each of these reports needs to be triaged. An important step in the triage process is the assignment of the report to a developer. Most previous efforts towards improving bug report assignment have focused on using an activity-based approach. We address some of the limitations of activity-based approaches by proposing a two-phased location-based approach where bug report assignment recommendations are based on the predicted location of the bug. The proposed approach utilizes a noun extraction process on several information sources to determine bug location information and a simple term weighting scheme to provide a bug report assignment recommendation. We found that by using a location-based approach, we achieved an accuracy of 89.41% and 59.76% when recommending five developers for the Eclipse and Mozilla projects, respectively.

A time-based approach to automatic bug report assignment

Improve automatic bug assignment (ABA) accuracy by using metadata in term weighting.Improve accuracy of common term-weighting technique, tf-idf, up to 14%.Recommend a light method for ABA based on the new term-weighting technique.Outperform the ML and IR methods by recommended method up to 55%. Bug assignment is one of the important activities in bug triaging that aims to assign bugs to the appropriate developers for fixing. Many recommended automatic bug assignment approaches are based on text analysis methods such as machine learning and information retrieval methods. Most of these approaches use term-weighting techniques, such as term frequency-inverse document frequency (tf-idf), to determine the value of terms. However, the existing term-weighting techniques only deal with frequency of terms without considering the metadata associated with the terms that exist in software repositories. This paper aims to improve automatic bug assignment by using time-metadata in tf-idf (Time-tf-idf). In the Time-tf-idf technique, the recency of using the term by the developer is considered in determining the values of the developer expertise. An evaluation of the recommended automatic bug assignment approach that uses Time-tf-idf, called ABA-Time-tf-idf, was conducted on three open-source projects. The evaluation shows accuracy and mean reciprocal rank (MRR) improvements of up to 11.8% and 8.94%, respectively, in comparison to the use of tf-idf. Moreover, the ABA-Time-tf-idf approach outperforms the accuracy and MRR of commonly used approaches in automatic bug assignment by up to 45.52% and 55.54%, respectively. Consequently, consideration of time-metadata in term weighting reasonably leads to improvements in automatic bug assignment.

An Effective Recommender Algorithm for Cold-Start Problem in Academic Social Networks

Abundance of information in recent years has become a serious challenge for web users. Recommender systems (RSs) have been often utilized to alleviate this issue. RSs prune large information spaces to recommend the most relevant items to users by considering their preferences. Nonetheless, in situations where users or items have few opinions, the recommendations cannot be made properly. This notable shortcoming in practical RSs is called cold-start problem. In the present study, we propose a novel approach to address this problem by incorporating social networking features. Coined as enhanced content-based algorithm using social networking (ECSN), the proposed algorithm considers the submitted ratings of faculty mates and friends besides user’s own preferences. The effectiveness of ECSN algorithm was evaluated by implementing it in MyExpert, a newly designed academic social network (ASN) for academics in Malaysia. Real feedbacks from live interactions of MyExpert users with the recommended items are recorded for 12 consecutive weeks in which four different algorithms, namely, random, collaborative, content-based, and ECSN were applied every three weeks. The empirical results show significant performance of ECSN in mitigating the cold-start problem besides improving the prediction accuracy of recommendations when compared with other studied recommender algorithms.

Automatic Bug Assignment Using Information Extraction Methods

The number of reported bugs in large open source projects is high and triaging these bugs is an important issue in software maintenance. As a step in the bug triaging process, assigning a new bug to the most appropriate developer to fix it, is not only a time-consuming and tedious task. The triager, the person who considers a bug and assigns it to a developer, also needs to be aware of developer activities at different parts of the project. It is clear that only a few developers have this ability to carry out this step of bug triaging. The main goal of this paper is to suggest a new approach to the process of performing automatic bug assignment. The information needed to select the best developers to fix a new bug report is extracted from the version control repository of the project. Unlike all the previous suggested approaches which used Machine Learning and Information Retrieval methods, this research employs the Information Extraction (IE) methods to extract the information from the software repositories. The proposed approach does not use the information of the bug repository to make decisions about bugs in order to obtain better results on projects which do not have many fixed bugs. The aim of this research is to recommend the actual fixers of the bugs. Using this approach, we achieved 62%, 43% and 41% accuracies on Eclipse, Mozilla and Gnome projects, respectively.

Focus Group Discussion Model for Requirements Elicitation Activity

Reviewing existing requirements elicitation models lead to the identification of an environment, detailed activities and tool support are very significant in understanding requirements elicitation activity. The environment ensures involvement from all stakeholders during the requirements elicitation activity in one platform, while the activities are steps that are very important in achieving the goal of requirements elicitation. The support component employs suitable technique that provides platform to groups of stakeholders to participate in requirements elicitation. Emphasizing these three components, in which focus group discussion for requirements elicitation (FGDRE) gives an understanding about the requirements elicitation activity and recommends essential requirements for requirements elicitation tool.

E-cost estimation using expert judgment and COCOMO II

The accuracy and efficiency of cost estimation methodology for web-based application is very important for software development as it would be able to assist the management team to estimate the cost. Furthermore, it will ensure that the development of cost is within the planned budget and provides a fundamental motivation towards the development of web-based application project. The literature review reveals that COCOMO II provides accurate result because more variables are considered including reuse parameter. The parameter is one of the essential variables in estimating the cost in web-based application development. This research investigates the feasibility to combine and implement COCOMO II and expert judgment technique in a tool called WebCost. In estimating a cost, the tool considers all variables in COCOMO II and requires expert judgment to key-in the input of the variables such as project size, project type, cost adjustment factor and cost driven factor. Developed in JAVA, WebCost is proven able to estimate cost and generate its estimation result. The usability evaluation conducted had shown that WebCost is usable when compared with other tools, it has its own advantages. WebCost is evidence suitable for everyone especially the project managers, software practitioners or software engineering student in handling the cost estimation tasks.

Developing tool for crosscutting concern identification using NLP

Aspect-oriented requirements engineering shows encouraging results in improving, identification, modularization and composition of crosscutting concerns. For instance, many techniques have been developed for crosscutting concern identification activity. This crucial activity is useful at the requirements level; it avoids coupling between requirements, improves traceability among requirements, eases function modularization, reduces software complexity, enhances the correctness of the software design, and most importantly it saves the cost. However, existing studies do not provide us with a clear definition of the term crosscutting concern and aspect. In terms of tools, most of these techniques in the existing studies are only supported by semi-automated tool. It means human intervention is required to achieve the desired results. In this paper, we describe a tool that provides automated support for identifying crosscutting concern at the requirements level. The tool utilises natural language processing technique to reason about properties of the concerns and model their structure and relationship.

Extracting features from online software reviews to aid requirements reuse

Display Omitted The extraction of software features from Software Requirement Specifications (SRS) is viable only to practitioners who have the access.Online reviews for software products can be used as input for features extraction to assist requirements reuse.Techniques from unsupervised learning and Natural Language Processing is employed as a propose solutions to Requirements Reuse problem.The approach obtained a precision of 87% (62% average) and a recall of 86% (82% average), when evaluated against the truth data set created manually. Sets of common features are essential assets to be reused in fulfilling specific needs in software product line methodology. In Requirements Reuse (RR), the extraction of software features from Software Requirement Specifications (SRS) is viable only to practitioners who have access to these software artefacts. Due to organisational privacy, SRS are always kept confidential and not easily available to the public. As alternatives, researchers opted to use the publicly available software descriptions such as product brochures and online software descriptions to identify potential software features to initiate the RR process. The aim of this paper is to propose a semi-automated approach, known as Feature Extraction for Reuse of Natural Language requirements (FENL), to extract phrases that can represent software features from software reviews in the absence of SRS as a way to initiate the RR process. FENL is composed of four stages, which depend on keyword occurrences from several combinations of nouns, verbs, and/or adjectives. In the experiment conducted, phrases that could reflect software features, which reside within online software reviews were extracted by utilising the techniques from information retrieval (IR) area. As a way to demonstrate the feature groupings phase, a semi-automated approach to group the extracted features were then conducted with the assistance of a modified word overlap algorithm. As for the evaluation, the proposed extraction approach is evaluated through experiments against the truth data set created manually. The performance results obtained from the feature extraction phase indicates that the proposed approach performed comparably with related works in terms of recall, precision, and F-Measure.

Using Ethnography Techniques in Developing a Mobile Tool for Requirements Elicitation

Requirements Elicitation is the process of discovering, extracting and gathering requirements for a system through communication with the stakeholders. Very often the process is carried out by observing the workplace and analyzing their needs at the work context. Ethnography, it is an elicitation technique that supports context-aware requirements elicitation. Common tools used in requirements elicitation process are mostly desktop-based which decrease the mobility features of bringing them to workplace. Recent advances in mobile computing technologies intend that mobile devices have the potential to support elicitation activities in the workplace. The aim of this research is to develop a mobile tool to support ethnography-based techniques to help the analysts directly observe the stakeholders need and help them in better understanding the system requirements. This paper describes the ongoing research and presents the progress so far.

L3OP: Learning Styles Application Using Learning Objects Approach

Learning Objects is not a new approach in electronic learning system. It has been applied since more than seven years ago. It can be clearly seen in CanCore project [9] in Canada and the emerging of several standards of Learning Objects by well-known organizations such as IEEE. L3OP (Learning Objects Technology in Object-Oriented Programming ELearning System) project applies the Learning Objects concept into electronic learning system focusing on digitalizing learning styles (especially among higher education students). The electronic learning systems have been used by these students since past few generations and many of them have been enhanced. However, some students still feel that the electronic learning systems are not as effective as they may seem. The reason is varieties of traditional methods of learning styles used by these students, for example, jotting down and highlighting notes cannot be done in the electronic learning systems. For example, some students prefer to use highlight pen or marker to underline the important notes. However, the electronic learning systems can not digitize the styles. The project is only concentrated on visual type of learners. The Learning Objects approach used in this project not only to ensure the organization of the database but also the reusability of the learning content itself.