Our research examines how cognitive tools, computational algorithms, and interactive media can improve learning outcomes and learner engagement. We aim to invent successful technologies, processes, and know-how that can be replicated on a wider scale. Our close partnership with the Institute of Cognitive Science (ICS) at the University of Colorado brings expertise and leading research in fields critical to educational innovation, including computer science, cognitive psychology, educational theory, and computational linguistics. In collaboration with ICS and other University partners, we engage in both use-inspired basic research and applied research. Use-inspired basic research extends the frontiers of knowledge in ways that are motivated by considerations of use and human or societal need.
Three research projects are examining how machine learning and natural language processing algorithms can contribute to creating engaging, personalized learning environments (CLICK), and supporting scalable curation of content in large-scale digital repositories (QUALITY and Educational Standards Alignment). A fourth project is investigating how curriculum adaptation tools can contribute to the development of teacher professional knowledge and sustainable curriculum (Teaching Box Builder).
Customized Learning Service for Concept Knowledge (CLICK)
A key educational finding from learning research is that every student brings preconceptions about how the world works to every learning situation, and that these initial understandings need to be explicitly targeted as part of an effective instructional process. This research, funded by the Advanced Learning Technologies program of the National Science Foundation, is designing and evaluating an end-to-end prototype of a “customized learning service for concept knowledge”. This software service will support learning environments to perform customizations based on a real-time analysis of what students understand about a particular topic. Learner-centered customizations are performed by algorithmically comparing a learner’s current conceptual understanding, depicted as a concept map, with a domain competency model generated automatically from selected digital library resources. These comparisons will enable learning environments to provide customized retrieval, delivery, and presentation of educational resources drawn from digital libraries. In computer science, this research is contributing to the development of multi-document summarization techniques capable of analyzing and synthesizing educational resources. In learning science, this research is contributing to our understanding of how common student misconceptions can be reliably identified by computer algorithms capable of analyzing student work.
Selected Publications about CLICK:
Ahmad, F., de la Chica, S., Butcher, K., Sumner, T. and Martin, J.H. Towards automatic conceptual personalization tools. In Proceedings of the 7th ACM/IEEE-CS Joint Conference on Digital Libraries, (Vancouver, British Columbia, Canada, 2007), accepted for publication. Download pdf
de la Chica, S. SciNews Online: Scaffolding the construction of scientific explanations. In Proceedings of the 2007 SIGCHI Conference on Human Factors in Computing Systems, Extended Abstracts - ACM Student Research Competition, (San Jose, California, 2007), 2183-2188. Download pdf
de la Chica, S., Ahmad, F., Martin, J.H. and Sumner, T.R. Supporting science understanding through a customized learning service for concept knowledge. In Proceedings of the Combined Workshop on Language-Enabled Educational Technology and Development and Evaluation of Robust Spoken Dialogue Systems, 17th European Conference on Artificial Intelligence, (Riva del Garda, Italy, 2006), 9-16. Download pdf
Developing a Computational Model of Quality for Digital Repositories (QUALITY)
Digital resource quality has emerged as a dominant yet poorly understood concern within digital library efforts and other content repositories, particularly those supporting community and user contributions. Evaluating resource quality involves making complex, time-consuming, and variable human judgments. Developing computational models of quality that approximate expert human judgments is a foundational requirement for developing interfaces and tools that can optimize and scaffold human judgments on quality. This research, funded by the Information and Intelligent Systems program at the National Science Foundation, is investigating the characteristics of digital resources that serve as key markers of quality for experts engaged in resource selection and collection curation, and determining how to model these markers computationally. Additionally, this research is assessing how machine learning and natural language processing techniques can be applied to recognize these markers with sufficient discrimination to approximate and scaffold effective human-decision making.
Selected Publications about QUALITY:
Custard, Myra and Tamara Sumner. (2005). Using machine learning to support quality judgements. D-Lib Magazine, 11(10). Download article
Educational Standards
Digital Learning Sciences is working with the Center for Natural Language Processing (CNLP) at Syracuse University to develop and evaluate tools and algorithms to support the automatic identification of relevant K-12 science and math education standards for a given digital learning resource. CNLP is developing algorithms that use natural language processing and machine learning techniques to analyze the content of a digital resource and suggest the most likely standards that are correlated with the content. Digital Learning Sciences is evaluating the performance of these algorithms and developing tools and interfaces to integrate them into cataloging and collection management workflows. This project is also creating algorithms for automatically aligning state educational standards to selected national standards to support state to state correlations. Outcomes from this project will improve both the ability of teachers to locate resources in the National Science Digital Library that support standards–based instruction and the ability of library builders to efficiently develop collections that support standards-based education.
Selected Publications about Educational Standards Alignment:
Devaul, Holly, Anne Diekema, and Jonathan Ostwald. (2007). Computer-Assisted Assignment of Educational Standards Using Natural Language Processing. Unpublished technical report. Boulder, CO.
Diekema, Anne and Holly Devaul. (2006). Computer Assisted Standard Assignment and Alignment. On NSDL ToolTime series [ReadyTalk demonstration]. Boulder, CO: National Science Digital Library (NSDL) retrieved April 14, 2006. Download demonstration
Devaul, Holly and Karon Kelly. (2003). Searching by educational standards in Digital Library for Earth System Education (DLESE): What does it mean and what do users want? (Poster presentation). 2003 National Science Digital Library Annual Meeting; National Science Digital Library (NSDL). Retrieved April 14, 2006. Download pdf
Teaching Box Builder
Educational digital libraries such as the Digital Library for Earth System Education (www.dlese.org) offer a multitude of pedagogically innovative digital resources for science education. Educators require pedagogical support to select and reuse these resources in their classroom teaching. This research is investigating how supporting teachers’ adaptation of curriculum can facilitate effective resource reuse and the development of sustainable curriculum based on digital library resources. One outcome of this research is the ENACT conceptual model, which articulates the relationships between the development of teacher professional knowledge and different forms of curriculum adaptation. This model is based on educational theories of learning and teacher knowledge, as well as interviews with educators and empirical studies of successful adaptation. Another outcome of this research is the Teaching Box Builder, which uses the Fedora digital repository environment to integrate curriculum content, educational standards, learning progressions, and digital library resources drawn from DLESE. The Teaching Box Builder is designed to support the adaptation processes specified in the ENACT conceptual framework and will be evaluated with practicing science educators.
Selected Publications about Teaching Box Builder:
Khan, Huda and Keith Maull. (2007). Realizing the role of digital repositories in educational applications: Supporting content and context. Second International Conference on Open Repositories: San Antonio, TX. Download pdf
Khan, Huda and Keith E. Maull. (2006). Teaching boxes: Customizable contextualized digital resources. ED-MEDIA 2006 World Conference on Educational Multimedia, Hypermedia and Telecommunications: Orlando, FL, p. 11-39, (June 26-30, 2006). Download pdf
Khan, Huda and Keith E. Maull. (2006). Teaching box builder: Customizing pedagogical contexts for use of digital library resources in classrooms. Poster presentation. 6th ACM/IEEE-CS Joint Conference on Digital Libraries (JCDL 2006): Chapel Hill, NC, (June 11-15, 2006). Download pdf
