Dominoes: An Interactive Exploratory Data Analysis Tool for Software Relationships

Project comprehension questions, such as "which modified artifacts can affect my work?" and "how can I identify the developers who should be assigned to a given task?" are difficult to answer, require an analysis of the project and its data, are context specific, and cannot alway...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on software engineering 2022-02, Vol.48 (2), p.377-396
Hauptverfasser: da Silva Junior, Jose Ricardo, Campagna, Daniel Prett, Clua, Esteban, Sarma, Anita, Murta, Leonardo
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Project comprehension questions, such as "which modified artifacts can affect my work?" and "how can I identify the developers who should be assigned to a given task?" are difficult to answer, require an analysis of the project and its data, are context specific, and cannot always be pre-defined. Current research approaches are restricted to post hoc analyses over software repositories. Very few interactive exploratory tools exist since the large amount of data that need to be analyzed prohibits its exploration at interactive rates. Moreover, such analyses typically require the user to create complex scripts or queries to extract the desired information from data. Here we present Dominoes, a tool for interactive data exploration aimed at end users (i.e., project managers or developers). Dominoes allows users to interact with different types and units of data to investigate project relationships and view intermediate results as charts, tables, and graphs. Additionally, it allows users to save the derived data as well as their exploration paths for later use. In a scenario-based evaluation study, participants achieved a success rate of 86 percent in their explorations, with a mean time of 7.25 minutes for answering a set of (project) exploration questions.
ISSN:0098-5589
1939-3520
DOI:10.1109/TSE.2020.2988241