Machine Learning Analysis for Metaphors Related User Interface Elements

Abstract

The user interface is a systematic representation of various controls and components. These controls are referred to as Metaphors in human-computer interaction. After software development, the functionality of software is improved by various updates. However, the software interface is improved by usability engineering. Metaphors play an important role in software accessibility. There is a wide range of user feedback in the form of textual comments over websites, blogs, and forums. In this study, we have performed short text analysis over these user comments. The text extraction is made by labeling text with keywords. After getting keywords, we have applied clustering to collect the clusters representing a specific Metaphor with a unique cluster. These clusters are also verified in search engines. Then we have compared Metaphors of smart devices with the general UI of personal computers. The mechanism presented serves as the short text analysis-based user interface analysis. The SOM graphs are explained to highlight the better visualization of Metaphors in two major computing devices as mentioned above. Clustering grouped the UI elements into six metaphor themes, and the shopping-cart metaphor reached 86% classification accuracy, showing that user comments can guide designers in choosing metaphors for both PCs and smartphones.