User Interface Evaluation with Machine Learning Methods

Show simple item record

dc.contributor.author Mao, Yanxun
dc.date.accessioned 2019-07-08T19:44:50Z
dc.date.available NO_RESTRICTION
dc.date.available 2019-07-08T19:44:50Z
dc.date.issued 2019
dc.date.submitted 2019
dc.identifier.uri http://hdl.handle.net/2027.42/149942
dc.description.abstract With the increasing complexity of user interfaces and the importance for usability evaluation, efficient methods for evaluating the usability of user interfaces are needed. Through this dissertation research, two computational models built with machine learning methods are introduced to evaluate user interface usability. This research consists of two phases. Phase I of the research implements the method of support vector machine to evaluate usability from static features of a user interface such as widget layout and dimensions. Phase II of the research implements the method of deep Q network to evaluate usability from dynamic features of a user interface such as interaction performance and task completion time. Based on the research results, a well-trained Phase I model can distinguish and classify user interfaces with common usability issues and is expected to recognize those issues when sufficient data is provided. Phase II model can simulate human-interface interaction and generate useful interaction performance data as the basis for usability analysis. The two phases of the research aim to overcome the limitations of traditional usability evaluation methods of being time-consuming and expensive, and thus have both practical and scientific values. From the practical perspective, this research aims to help evaluate and design user interfaces of computer- based information systems. For example, today’s application software development on computer based information system always integrates many functions or task components into one user interface page. This function integration needs to be carefully evaluated to avoid usability issues and the competitive field of software development requires an evaluation process with short cycles. Phase I and Phase II of the research provide an efficient but not necessarily comprehensive usability evaluation tool to meet some of the demands of the field. From the scientific perspective, this research aims to help researchers make quantifiable predictions and evaluations of user interfaces. Qualitative theories and models are important, but often insufficient for rigorous understanding and quantitative analysis. Therefore, this research work on computational model-based interface evaluation has important theoretical value in advancing the science of studying human behavior in complex human-machine-environment systems.
dc.language.iso en_US
dc.subject user interface
dc.subject usability evaluation
dc.subject machine learning method
dc.title User Interface Evaluation with Machine Learning Methods
dc.type Thesis
dc.description.thesisdegreename PHD
dc.description.thesisdegreediscipline Industrial & Operations Engineering
dc.description.thesisdegreegrantor University of Michigan, Horace H. Rackham School of Graduate Studies
dc.contributor.committeemember Liu, Yili
dc.contributor.committeemember Burns Jr, Daniel M
dc.contributor.committeemember D'Souza, Clive Rahul
dc.contributor.committeemember Yang, Xi (Jessie)
dc.subject.hlbsecondlevel Industrial and Operations Engineering
dc.subject.hlbtoplevel Engineering
dc.description.bitstreamurl https://deepblue.lib.umich.edu/bitstream/2027.42/149942/1/myx_1.pdf
dc.identifier.orcid 0000-0002-7968-7110
dc.identifier.name-orcid Mao, Yanxun; 0000-0002-7968-7110 en_US
dc.owningcollname Dissertations and Theses (Ph.D. and Master's)
 Show simple item record

This item appears in the following Collection(s)


Search Deep Blue

Advanced Search

Browse by

My Account

Information

Available Now


MLibrary logo