Date of Publication

2-8-2021

Document Type

Master's Thesis

Degree Name

Master of Science in Computer Science

Subject Categories

Computer Sciences

College

College of Computer Studies

Department/Unit

Computer Science

Thesis Advisor

Macario O. Cordel, II

Defense Panel Chair

Rafael A. Cabredo

Defense Panel Member

Toni-Jan Monserrat
Conrado D. Ruiz, Jr

Abstract/Summary

Rapid prototyping is a process used in mobile application development, and several studies have attempted to automate some parts of the rapid prototyping process. Nonetheless, these studies focused on (1) wireframe generation and (2) translation of wireframes to code. In this work, rather than focusing on these two well-studied rapid prototyping processes, we aim to investigate automating the wireflow organization task using machine learning techniques. This work consists of several parts that are components of wireflow organization. A dataset was first built composed of 754 annotated wireflow samples. The dataset consists of 10,994 mobile UI images with 2,300 annotated interaction elements. Experiments on machine learning (ML) models were conducted and evaluated to produce a potential classifier to predict the next wireframe. This first study on wireflow prediction shows that the tree-based ML models performed significantly better than non-tree based ML models. This work also explored supplementary classifiers for interaction element detection and wireframe classification. These classifiers produced results with varying significance and the possibility of an end-to-end wireflow prediction model.

Abstract Format

html

Language

English

Format

Electronic

Physical Description

167 leaves

Keywords

User interfaces (Computer systems)--Design; Rapid prototyping; Machine learning; Forecasting; Mobile apps

Upload Full Text

wf_yes

Embargo Period

5-19-2022

Share

COinS