Exploring the Content Ecosystem of the First Open-source Adaptive Tutor and its Applications on Intelligent Textbooks
- 1. Exploring the Content Ecosystem of the First Open-source Adaptive Tutor and its Applications on Intelligent Textbooks Ioannis Anastasopoulos, UC Berkeley School of Education Fifth Workshop on Intelligent Textbooks (iTextbooks) ioannisa@berkeley.edu
- 2. Introduction ● Intelligent tutoring systems and other adaptive tutors popularized over past few decades - content environments of tutoring systems are often overlooked. ● Efforts over the past years to incorporate adaptive and interactive content/activities into digital textbooks. ● OATutor successfully creates a sustainable content environment, while fostering a mastery learning algorithm. ● The OATutor content environment is replicable, and can facilitate the creation of content that can be easily integrated into digital textbooks.
- 3. Related Work (Content at Scale) ● Scaling content usually requires large amounts of funding (Khan Academy, MATHia, ASSISTments) ○ Not easy to sustain in an open-source environment ● Open Educational Resources partially circumvent this issue through user-generated content collection, but still require forms of financial support ○ Such compilation of content also requires community to support to promote high-quality content ● Difficulty balancing user-friendly content creation environments, and potential to create complex problems
- 4. Related Work (Crowdsourcing/Learnsourcing) ● To counter financial difficulties many systems turn to crowdsourcing ○ Issue of maintaining quality content ● Learnersourcing places the learner in the places of the crowd ○ Help with higher quality/relevant content ○ Generation effect provides benefits for the learner
- 5. Related Work (Advances in Intelligent textbooks) ● Attempts to explore the usage of adaptive content presentation and content recommendation within textbooks ● Tools developed to facilitate combination of adaptive textbooks and interactive content ● Struggle of content generation, resulting in push for automated content mapping
- 6. Introduction to OATutor ● Open Adaptive Tutor (OATutor) was built upon ITS principles to facilitate mastery learning content (open-source). ● Offers creative commons licensed content, with personalized help for each problem. ● Questions curated from OpenStax textbooks with help built by the OATutor content team. https://github.com/CAHLR/OATutor
- 9. Content From OpenStax ● OpenStax was selected from a collection of OERs due to its creative commons license and collection of content. ● The OATutor content team transcribed three algebra textbooks, and one statistics textbook, from OpenStax into tutor lessons. ● 15-30 problems generated per OpenStax lesson. ● OATutotr spreadsheet content structure mimics that of a textbook question with added features for help capabilities.
- 10. Example problem from OATutor Content Spreadsheet
- 11. OATutor Content Creation ● Includes knowledge component column to identify correct KC ● Directly curated from OpenStax problems, with hints created by the content team ● Automated script converts the problems created on the spreadsheet into problems in the system’s frontend ● Spreadsheet format allows to provide simple interface, but with potential to be modified for complex problem creation
- 13. Applications on Intelligent Textbooks Creating a Bridge Between Digital Textbooks and Adaptive Tutors
- 14. Content Creation Connection to Textbooks ● Adaptive lessons can be used to supplement a digital textbook ● Feasibility of learning-objective focused content generation directly from digital textbooks ○ Facilitated through the crowdsourced community of learners ■ Researcher apprentice programs such as UC Berkeley’s URAP provide the benefits of Learnersourcing, and also overcome the issue of having an expert learner ■ Avoids the issue of funding for generating content at scale ○ Hint generation based on OpenStax lessons to go hand-in-hand with the textbook
- 15. OpenStax Integration ● OATutor lessons have a 1-1 correspondence with OpenStax lessons ○ Additionally, the problems themselves correspond to OpenStax problems ○ Pilot tested in community college classroom that utilizes OpenStax ● Possible to integrate OATutor directly into OpenStax, embedding it within the question section of each chapter ● Such a process could be generalized to other textbooks, by having content curated for them in OATutor, and embedding the adaptive tutor into the digital textbook
- 16. Alternative Integration ● Individual problems within textbook directly link to tutor question corresponding to them ○ This would allow for OATutor to track learner mastery as they go through the section ● Future questions would be grayed out/optional if the learner has already achieved mastery ● These elements would be included in a subtle way to avoid altering a pdf/printed version of the textbook ○ Potentially appearing upon learner interaction with the digital version
- 17. Other Potential Implementations ● Incorporation of AI tools (such as ChatGPT) into the hint generation process to speed up content curation ○ Such tools could also be used to attempt to create hints in a similar language to that of a particular textbook ● Allowing the pairing of homework/assessment style content directly within the textbook ○ Homework-style content could receive appropriate hints/scaffolds (or have none in the case of a mastery assessment) ○ Additional pilots of the system would be required to explore such ideas
- 18. Concluding Thoughts ● Crowdsourced community of learners helps overcome major challenges of creating content at scale ● OATutor content is drawn directly from textbooks, and the way content is created allows for a 1-1 mapping with textbook problems ● This type of content ecosystem enables the potential to connect digital textbook to adaptive tutor content and capabilities ● The open-source nature of OATutor allows for future work and experimentation with its integration into intelligent textbooks
- 19. Thank you! Contact email: ioannisa@berkely.edu Questions?