Matching Items (15)
Filtering by
- Creators: Arizona State University
- Creators: Department of Information Systems
- Creators: Armfield, Jessica Ann
Description
In natural language processing, language models have achieved remarkable success over the last few years. The Transformers are at the core of most of these models. Their success can be mainly attributed to an enormous amount of curated data they are trained on. Even though such language models are trained on massive curated data, they often need specific extracted knowledge to understand better and reason. This is because often relevant knowledge may be implicit or missing, which hampers machine reasoning. Apart from that, manual knowledge curation is time-consuming and erroneous. Hence, finding fast and effective methods to extract such knowledge from data is important for improving language models. This leads to finding ideal ways to utilize such knowledge by incorporating them into language models. Successful knowledge extraction and integration lead to an important question of knowledge evaluation of such models by developing tools or introducing challenging test suites to learn about their limitations and improve them further. So to improve the transformer-based models, understanding the role of knowledge becomes important. In the pursuit to improve language models with knowledge, in this dissertation I study three broad research directions spanning across the natural language, biomedical and cybersecurity domains: (1) Knowledge Extraction (KX) - How can transformer-based language models be leveraged to extract knowledge from data? (2) Knowledge Integration (KI) - How can such specific knowledge be used to improve such models? (3) Knowledge Evaluation (KE) - How can language models be evaluated for specific skills and understand their limitations?
I propose methods to extract explicit textual, implicit structural, missing textual, and missing structural knowledge from natural language and binary programs using transformer-based language models. I develop ways to improve the language model’s multi-step and commonsense reasoning abilities using external knowledge. Finally, I develop challenging datasets which assess their numerical reasoning skills in both in-domain and out-of-domain settings.
ContributorsPal, Kuntal Kumar (Author) / Baral, Chitta (Thesis advisor) / Wang, Ruoyu (Committee member) / Blanco, Eduardo (Committee member) / Yang, Yezhou (Committee member) / Arizona State University (Publisher)
Created2023
Description
Language Textbooks often play a major role in English for speakers of other languages (ESOL) classrooms, not only to provide information but also to help students construct their identities. Several studies showed that identity and language learning are inseparable, due to the simple fact that every learner has his/her own identity aspects like race, gender, social class, and speaker status (i.e., native speaker vs. non-native speaker of the target language). These aspects should be acknowledged because providing the students with limited identity options might cause the students to resist learning the language or be less invested in this practice (Norton & Toohey,2011).
However, there is limited support for teachers who wish to examine identity in ESOL textbooks. Several scholars attempted to evaluate the range of identity options offered in ESOL textbooks, but they all used either Critical Discourse Analysis or Content Analysis which can be effective; however, these procedures require training and can take a long time, so they may not be practical for teachers. This suggests that there is a need for a less complicated evaluation tool that can be easily used by teachers.
The purpose of this thesis is to develop a teacher-friendly identity-focused checklist for ESOL textbooks, and the thesis is guided by the following questions: (a) what would an evaluation checklist for identity in ESOL textbooks look like?; (b) what can this checklist reveal about ESOL textbooks? The purpose of this thesis was achieved by developing a qualitative checklist that covers, race, gender, social class, and speaker status, and demonstrating how to use it on a collection of five adult ESOL textbooks. The checklist revealed similarities and differences between the textbooks, including important shortcomings, and that kind of information can be useful for the teacher to make decisions about the textbook he/she uses.
However, there is limited support for teachers who wish to examine identity in ESOL textbooks. Several scholars attempted to evaluate the range of identity options offered in ESOL textbooks, but they all used either Critical Discourse Analysis or Content Analysis which can be effective; however, these procedures require training and can take a long time, so they may not be practical for teachers. This suggests that there is a need for a less complicated evaluation tool that can be easily used by teachers.
The purpose of this thesis is to develop a teacher-friendly identity-focused checklist for ESOL textbooks, and the thesis is guided by the following questions: (a) what would an evaluation checklist for identity in ESOL textbooks look like?; (b) what can this checklist reveal about ESOL textbooks? The purpose of this thesis was achieved by developing a qualitative checklist that covers, race, gender, social class, and speaker status, and demonstrating how to use it on a collection of five adult ESOL textbooks. The checklist revealed similarities and differences between the textbooks, including important shortcomings, and that kind of information can be useful for the teacher to make decisions about the textbook he/she uses.
ContributorsAlmatard, Abdulmajeed Mohammed (Author) / James, Mark (Thesis advisor) / Matsuda, Aya (Committee member) / Prior, Matthew (Committee member) / Arizona State University (Publisher)
Created2019
Description
The purpose of this study was to increase microlearning training module usage and completions by 10–15% over a 30-day period by including evaluation in the design and development of a new microlearning training module in the golf equipment industry. Evaluation was conducted using a bespoke evaluation tool, which was designed and developed using design thinking methodology. The evaluation tool was applied to two previously designed microlearning modules, Driver Distance B and Driver Distance C, both of which served as comparisons for the new module’s completion data. Evaluation reports were generated that informed the development of the new module, named Golf Software. This action research study was grounded in constructivist learning theory, design thinking, and dashboards research. A nested, case study-mixed methods (CS- MM) design and a sequential qualitative to quantitative design were used. Research was conducted with the Knowledge Management Department at Ping, an original golf equipment manufacturer (OEM) in Phoenix, Arizona. Participants included three eLearning Designers, which included the researcher as a participant observer. Qualitative data included interviews, reflective researcher journal, and artifacts such as the new microlearning training module and evaluation reports. Quantitative data included completion numbers collected from the organization’s learning management system (LMS) and email campaign service. Findings from this study were mixed, with the new module’s completion numbers 20.27% greater than Driver Distance C and 7.46% lower than the Driver Distance B. The objective of this study was not met, but outcomes provided valuable information about incorporating evaluation in the Knowledge Management Department’s instructional design process.
ContributorsRegan, Elizabeth (Author) / Marsh, Josephine P (Thesis advisor) / Leahy, Sean (Committee member) / Gretter, Sarah (Committee member) / Arizona State University (Publisher)
Created2020
Description
Making significant progress on the U.N. Sustainable Development Goals (SDGs) needs change agents equipped with key competencies in sustainability. While thousands of sustainability programs have emerged at various educational levels over the past decade, there is, as of yet, no reliable way to assess if these programs successfully convey key competencies in sustainability. This dissertation contributes to addressing this gap in three ways. First, it reviews the body of work on key competencies in sustainability. Based on broad agreement around five key competencies as well as an emerging set of three, an extended framework is outlined that can be used as unified set of learning objectives across sustainability programs. The next chapter reviews the scholarly work on assessing sustainability competencies. Based on this review, a typology of assessment tools is proposed offering guidance to both educators and researchers. Finally, drawing on experience of the four-year “Educating Future Change Agents” project, the last chapter explores the results from a diverse set of competency assessments in numerous courses. The study appraises assessment practices and results to demonstrate opportunities and challenges in the current state of assessing key competencies in sustainability. The results of this doctoral thesis are expected to make a practical and scholarly contribution to the teaching and learning in sustainability programs, in particular with regards to reliably assessing key competencies in sustainability.
ContributorsRedman, Aaron (Author) / Wiek, Arnim (Thesis advisor) / Barth, Matthias (Committee member) / Basile, George (Committee member) / Fischer, Daniel (Committee member) / Mochizuki, Yoko (Committee member) / Arizona State University (Publisher)
Created2020
Description
This research study aims to find out the way how goodwill should be evaluated. This paper is about accounting for goodwill which will provide general information about goodwill value, especially of public companies. Additionally, I will discuss sources of goodwill, the importance of goodwill, why it is important to evaluate goodwill correctly, and what methods have been applied to evaluate goodwill. This thesis will analyze the advantages and disadvantages of both methods of accounting for goodwill which are the impairment testing method and the amortization method. This study is done by researching studies, journal articles, reviews, books, and websites about accounting. Lastly, this study will provide a suggestion for how goodwill should be evaluated effectively.
ContributorsPham, Trang Thi Thuy (Author) / Shields, Paul (Thesis director) / Huang, Xiaochuan (Committee member) / School of Accountancy (Contributor) / Department of Information Systems (Contributor) / Barrett, The Honors College (Contributor)
Created2021-05