Enhancing Adaptive Courseware based on Natural Language Processing
Abstract
The proposed project to the Office of Naval Research (ONR) includes the development and deployment of an enhanced intelligent tutoring system with adaptive courseware and communication based on controlled natural language. It is a continuation of the project the Adaptive Courseware & Natural Language Tutor (AC&NL Tutor) intelligent tutoring system funded by the ONR under the Grant N00014-15-1-2789. The AC&NL Tutor is an environment for conceptual knowledge acquisition in the learning, teaching and knowledge testing processes. Our research focuses on improvements in courseware adaptivity and natural language communication in intelligent tutoring systems. The project outcomes include the development of an enriched and improved version of an almost fully automated intelligent tutoring system, which will be able to tutor any declarative domain knowledge and to communicate in natural language.This intelligent tutoring system will incorporate a hybrid design, allowing users to learn in an environment that is adaptive to their needs and knowledge. Such an intervention is needed because learners vary widely in their background knowledge and learning styles. Hence, the successful intelligent tutoring system will need to adapt and respond to the learners~ skills, knowledge, goals, motivation, and interests. To achieve this objective, we will build upon andadapt the existing AC&NL Tutor intelligent tutoring system. We will first examine the benefits and appeal of the existing system, and subsequently, modify it according to the research findings. To do so, we will conduct preliminary research to develop a better understanding of concept map mining. Various concept map mining techniques will be tested, and the most effective ones will be implemented. The domain knowledge will be enriched with hypermediaand cardinality. Consequently, questions generated during testing will support such enrichment of the domain knowledge. Moreover, testing items will be able to express questions in an objective form, i.e. multiple-choice, true-false, matching and completion. Based on individual learner performance, the use of a probability-based learner model will be investigated. Also, the most effective self-regulated learning strategies will be examined as well as the useful models that deal with specific learners~ behaviour while using intelligent tutoring systems (e.g. wheel spinning, gaming the system). In this way, both cognitive (intellectual capacity) and noncognitive skills (interest, responsibility, diligence) will be assessed, since understanding, reasoning and processing of information are important predictors of effective learner performance, but emotional intelligence should be considered as well. These efforts aim to lay the empirical foundations and to develop an effective system that will adapt to the wide variation in individual differences among learners.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 17, 2020
- Source ID
- N000142012066
Entities
People
- Ani Grubi
Organizations
- Office of Naval Research
- United States Navy
- University of Split