Application Screen Documentation
Overview
Dear Audio Recorder Developers,
We are a group of researchers from the College of William & Mary who are working on developing techniques for automatically documenting software. We have developed a new technique, called Clarity, that uses Deep Learning to automatically document the functionality of a given screenshot of an Android application. This model was trained on one of the largest known datasets of screenshots and corresponding captions that describe functionality. We have applied our model to document several screens from your application, which we have submitted (as html) as part of this pull/merge request. In addition, we have included the documentation in this comment, and this link to allow for quick inspection. This documentation was generated completely automatically with no human intervention. It includes a series of screenshots, the activity name of each screenshot, and either/both a general description of functionality, as well as a description of the functionality of individual UI components. If you think this documentation might be useful for your project or contributors, we hope that you will consider accepting this pull/merge request. In either case, we hope that you might have time to answer a few simple questions about the quality of automatically generated documentation included in this pull request:
Questions:
- Is the content of the captions included in the screenshots accurate and relevant for the given screenshot?
- Is there any extraneous information included in the captions? If so what?
- Are the captions readable and understandable (i.e., do they make adequate use of the English language)?
- If you were given the opportunity, how would you improve the generated captions (e.g., adding different types of functionality, more specific, less specific)?
- Would you include these captions as part of your app documentation (or manual)? Assuming that this documentation would be automatically generated for each app version.
Automated Documentation
