Exploranative Code Quality Documents

Abstract

A B C D Fig. 1. Exploranative code quality document for Lucene 2.0. A Textual overview in terms of quality attributes, code smells, and bugs, which includes embedded visualizations. B Overview visualizations: parallel coordinates plot and scatterplot. C Source code of a class provided in the details view. D Description of a quality attribute alternatively presented in the details view.Abstract-Good code quality is a prerequisite for efficiently developing maintainable software. In this paper, we present a novel approach to generate exploranative (explanatory and exploratory) data-driven documents that report code quality in an interactive, exploratory environment. We employ a template-based natural language generation method to create textual explanations about the code quality, dependent on data from software metrics. The interactive document is enriched by different kinds of visualization, including parallel coordinates plots and scatterplots for data exploration and graphics embedded into text. We devise an interaction model that allows users to explore code quality with consistent linking between text and visualizations; through integrated explanatory text, users are taught background knowledge about code quality aspects. Our approach to interactive documents was developed in a design study process that included software engineering and visual analytics experts. Although the solution is specific to the software engineering scenario, we discuss how the concept could generalize to multivariate data and report lessons learned in a broader scope.Manuscript received xx xxx. 201x; accepted xx xxx. 201x. Date of Publication xx xxx. 201x; date of current version xx xxx. 201x. For information on obtaining reprints of this article, please send e-mail to: reprints@ieee.org. Digital Object Identifier: xx.xxxx/TVCG.201x.xxxxxxx the workings of analysis algorithms will increase trust into the quality assessment. In contrast, some tools integrated into the software development chain already make a step in this direction, however, consider different quality aspects separately and use only simple visualizations. Our goal is to go a step ahead and integrate visualizations with contextual information. We argue that textual descriptions are key to communicate context and guide the data exploration process. Additional textual explanations can make the interface fully self-explanatory, allow the integration of abstract information that is hard to explain visually, and facilitate blending domain terminology with data descriptions.We suggest a visual analytics solution that combines text and visualization. The idea is to automatically generate textual descriptions that guide the users through the main findings of a code quality analysis. We augment the generated text with visualizations that also allow users to interactively explore different aspects of source code quality. With this, we adopt the ideas of exploranation (explanation and exploration) [58, 61] and explorable explanations [56] for a visual analytics arXiv:1907.11481v1 ...

A scalable "exploranation" technique for hierarchically indexed table data

“…Mumtaz and others stress the importance of consistent linking across the triad of data visualization, text, and word-sized charts embedded in the text [13]. From their argument, the SPIViewer can be improved upon in a few directions.…”

“…To compensate for the shortcoming of the difficulty of grasping the overview of the data through systems based on automated insight systems, Latif et al have proposed VIS Author Profiles for browsing co-authorship relationships of researchers, the interactive Map Reports for two-attribute datasets representing geographic distributions, and a visual analysis system for data assessed for readability and maintainability with respect to program code [8,9,13]. These systems attempt to explain the whole data rather than individual data facts using template based natural language processing technique.…”

“…One of the challenges for "exploranation (= exploration + explanation)" technology [13,21] is the scalability. The number of attributes in the table datasets treated in the past research studies was limited to a few.…”

Words of Estimative Correlation: Studying Verbalizations of Scatterplots

“…Study participants report the need to understand what the system meant when using words such as moderate and strong. Mumtaz et al [6] introduce a multimodal tool to help with code quality analysis and acknowledge this by including tooltips to clarify the boundaries when values change from low to medium. A slightly different approach, less focused on natural language, was used by Demiralp et al [5] when providing an insight-based visualization exploration system.…”

“…Recent advances in machine learning and artificial intelligence have paved the way for novel multimodal interactive methods that mix natural language and visualization to facilitate discovery and improve workflows within data analysis [1], [2], [3]. Natural language statements can be used to complement the information that is displayed visually, for example, by reducing clutter with statistical summaries [4], by providing complementary information in textual form that are not perceived easily with graphics [5], or as in the case of automated reports, by contextualizing visual summaries and charts within a narrative [6]. There is, however, a lack of empirical research on understanding how viewers verbalize their interpretations of visualizations and how they interpret verbalizations of visualizations in data analysis contexts.…”