Citation Context

...e ability of a developer to answer questions about a piece of code in a system was strongly determined by whether the developer had authored some of the code, and how much time was spent authoring it =-=[15]-=-. Mockus and Weiss used properties of an individual change to predict the probability of that change causing a failure [23]. They found that changes made by developers that were more experienced with ...

Citation Context

...t the assignment of developers to bug reports. It has been proposed that tasks such as bug triage can be improved if an externalized model of each programmer’s expertise of the code base is available =-=[12]-=-. Even though approaches for expertise models based on software repository contributions are available, existing recommendation systems for bug assignment typically use expertise models based on previ...

Citation Context

...estions: reasoning using facts they already know, communicating with teammates, or dynamically or statically exploring code. For code that developers know well, developers may already know the answer =-=[6]-=-. But this level of understanding is difficult to achieve due both to the number of reachability relationships present in a codebase and because they often change as developers edit the code. One fiel...

Citation Context

...ion data for determining expertise in or familiarity with source code. In a previous study, we considered whether interaction information alone could indicate for which code a developer had knowledge =-=[3]-=-. This study involved nineteen industrial Java programmers. Through this study, we found that DOI values computed from the interaction information can indicate knowledge about a program’s source. This...

Citation Context

... of the system increases as she works more on various components of the system [7, 39]. Knowledge gained from experience matters: it can lead to better ability to answer questions about previous work =-=[23]-=-, and better quality work [31]. These findings are also supported by Robillard [38], who suggests that lack of such knowledge could hurt software quality. We contribute to this literature in two ways....

Citation Context

...eaningfulness of navigation traces and their potential. Fritz et al. conducted an empirical study assessing the relationship between programmers activity and what a programmer knows about a code base =-=[13]-=- and DeLine et al. report results of two studies which demonstrate that sharing navigation data can improve program comprehension “and is subjectively preferred by users” [14]. Both Mylyn [15] and Nav...

Citation Context

...s of the project. Our approach uses several heuristic criteria to infer developer expertise, change activity, and commit contributions in the context of a particular file. According to a recent study =-=[5]-=-, the programmer activity (frequency and recent date) is a good indicator of the knowledge developers have in the code. We evaluate our approach on a number of open source projects. For each project, ...

Citation Context

...elopers first mentally construct a control flow representation of connections between its parts [20]. And their knowledge of a method’s part of the call graph increases as they interact with its code =-=[11]-=-. Information foraging theory predicts that developers traverse control flow and search for “prey” – locations in code – by using “scent” – the similarity of the information which labels the control f...

Citation Context

...ng their effects when controlling for code metrics known to have a relationship with failures: size, complexity, and churn. We use ownership as a proxy for expertise as followed by others [32], [33], =-=[34]-=- and evaluate the hypothesis that more changes by those with low expertise leads to more failures [35]. For each component we count the number of contributions and divide the proportion of total contr...