|dc.description.abstract||Three studies were conducted to explore machine learning support of novel learning problems. The first was a study of a computer-supported online learning website, which was designed to support teachers improving their pedagogy and this work created a data-first approach for Linguistic pedagogy, and to support improved pedagogy delivery and evaluation; the other two studies are supporting a participatory design project to create more youth-centered applications (i.e. an Android learning application), which is divided into two phases - the user interface information design phase to ascertain patterns in the data set, and phase two applied these pattern to support user sentiment analysis (i.e. collaborative and content-based filtering) to better support user experience in finding youth-centric content. The main objective of this work was to create frameworks that can be used to design usable applications and the proposed frameworks were evaluated through empirical experiments. We applied a framework for the design of online learning systems. This research will result in user sentiment analysis for generation z and other populations. Our goal was to validate the proposed framework and impact on designing online educational applications for these particular populations.
In the field of human-computer interaction, we work to define the process of working with humans to find better solutions to support their computer interactions through technology. HCI as defined "improves the design and uses of computer technology, and focused on the interfaces between people (users) and computers. Researchers in the field of HCI both observe the ways in which humans interact with computers and design technologies that let humans interact with computers in novel ways" (Card, Moran, & Newell, 1980). We have studied and worked with many communities in a process of participatory design and during the course of this work, we have gained a novel perspective on two communities of practice and their usage of computers and problems they may encounter. After studying these communities, we gathered functional requirements, created designs, and developed solutions to satisfy their requirements. In this manuscript, we will discuss work with two communities through three essays to document three studies. Our work was to support educational partners, gather requirements, and develop systems to better support their needs. In study one, we worked with teachers and found that they needed a better mechanism for returning results efficiently to their students. We studied the literature and found that the existing learning management systems are a great support to improve educational practice and provide an effective reinforcement that goes far beyond traditional classroom instruction (Ellis, 2009), but did not provide support for the needs of our teachers in the field of communications disorders. The faculty wanted to provide quicker and more robust feedback, to support an increase in student's confidence in learning in this problem area. Secondly, we aimed to provide support for teachers in this area, we endeavored to develop a smarter way to generate exams to optimize their usage of time to give them more time to devote to other higher-level instructional activities instead of spending great amounts of time on rote activities (i.e. grading by hand and creating exams). Recently there has been a resurgence of interest by academic institutions and companies, in the use of specialized algorithms to better solve problems. Some have emphasized an understanding of machine learning, the scientific study of algorithms and statistical models that computer systems use in order to perform a specific task effectively without using explicit instructions, relying on patterns and inference instead (Bishop, 2016). This field has been very beneficial to a multitude of systems where there is some amount of ambiguity. In our first study, we believed that the creation of specialized evaluations was an ambiguous (i.e. or not exact) process and planned to use techniques to generate smart evaluations through machine learning support.
In studies two and three, we worked with professionals from university extension and found that they needed a better user interface that was more supportive of younger students from generation z. Their existing web presence supports all age groups and though they do have some support for youth audiences, it is hard to discover on their web site (i.e. it is obscure and hard to navigate to). Their solution was to request the creation of a more youth-friendly website and with the prevalence of mobile technology use by youth (Druin, Bederson, Hourcade, Sherman, Revelle, Planter, & Weng, 2003), we were also requested to develop a mobile application to support this effort. We delivered a website and an Android-based mobile application to support our users' needs and requirements to provide more youth-friendly access. We created 3 versions of the mobile application and each mobile application had higher user acceptance rate as accessed through user studies and evaluation. Secondly, we aimed to provide support for users to find more relevant articles in their interest area, we endeavored to develop a smarter way to provide relevant materials instead of pre-determining our users' choices. Recently there has been great interest by companies and academic institutions in the use of crawler-based technologies and sentiment analysis to return better results (e.g. Amazon and Netflix). For example, as a person utilizing Amazon for online shopping and to procure an item, at the end of that experience the system will suggest another item to the customer that they think will match their interest or needs (i.e. sentiments). In Human-Computer Interaction, we can also refer to this an improving User Experience. Theoretically, this collaborative filtering (Pandey, 2019) or content-based filtering (Pandey, 2019) can be utilized to calculate group sentiment (e.g. sentiment analysis) and we perform this work to provide better results to our users based on their preferences to return better results. In machine learning, we create better methods by relying on patterns and inference and we utilized this method to improve results returned by our application (Bishop, 2016).||en_US