Adaptive Courseware based on Natural Language Processing (AC & NL Tutor)
Abstract
The performer proposes to develop and deploy a new model for an intelligent tutoring system with adaptive content and with communication based on controlled natural language. This project is a continuation of prior research in the field of using computers in education in the broadest sense, and of the development of intelligent tutoring systems. In this regard, since the beginning of 1992 the performer has been systematically exploring, developing, and applying: (i) a model of learning and teaching with computer teacher (TEx-Sys model) (Stankov, 1997; Stankov et al, 2008), (ii) processing controlled natural language over ontology (CoLaB Tutor) (Žitko, 2010), (iii) automatically and dynamically generated adaptive courseware (AC-ware Tutor) (Grubiši?, 2012). The TEx-Sys is a model of an intelligent hypermedia authoring shell for generating intelligent tutoring systems in freely chosen knowledge domains. This authoring shell is based on a number of concepts: the cybernetic model of systems, the paradigm "teaching is control of learning", the use of teaching principles of the "expert human tutors” with "one-to-one" tutoring model, in a traditional ITS architecture. Generating controlled language in the CoLaB Tutor enables the learner to learn using a sentence approach. This approach has enabled the implementation of one-way dialogue initiative in which a computer tutor asks questions, and the learner responds and gets feedback on a controlled language in a text-to-text manner. The Adaptive Courseware (AC) Tutor bases its knowledge on ontologically formalized domain knowledge and the adaptivity is realized using stereotypes and Bloom s knowledge taxonomy. Implemented prototype has automatic generation of courseware elements, dynamic selection and sequencing of courseware elements, automatic generation of test questions and initial test generated over a representative subset of the domain knowledge. The synergy of researchers’ experiences and all the mentioned models’ features will enable an environment for the acquisition of conceptual knowledge in the learning, teaching and knowledge testing process. Because of its fundamental features of adaptivity and natural language processing, the proposed project is called Adaptive Courseware & Natural Language Tutor (AC & NL Tutor). Project outcome includes a creation of almost fully automated intelligent tutoring system which will be able to tutor any declarative domain knowledge and to communicate in natural language. Automation will be reflected by adaption and generation of knowledge about the learner who is being taught and knowledge of methods for tutoring learners. The design of the domain knowledge in the expert module will also be automated (computer expert uses computer knowledge extraction), instead of manual (live expert designs domain knowledge). Moreover, the creation, selection, sequencing and presentation of courseware elements will be completely automated. Natural language processing will be applied during two-way communication in all learning, teaching and knowledge testing phases. Instruction design in this system will offer several scenarios of learning, teaching and testing learner’s knowledge.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2016
- Source ID
- N000141512789
Entities
People
- Ani Grubišić
Organizations
- Office of Naval Research
- United States Navy
- University of Split