Abstract:

Building intelligent computer programs supporting learning requires selection of an appropriate representation for knowledge and storing it in a database. Similarly, intelligent information retrieval and knowledge presentation systems require the application of databases. Embedding of knowledge structures into database structures makes possible efficient storing, indexing and data retrieval. This approach results in automating certain teaching tasks and in building solutions that scale up well as the number of students increases. It is also appropriate to the situations where the adjustment of learning processes to different learners' needs is crucial such as in the life-long learning.

The paper has three aims. The first is to verify to what degree simple, low-cost methods basing on an e-textbook partitioned into fragments and embedded in a database can help in building electronic tutors supporting student learning. At the beginning of our work we considered system Inquire (*) which was based on the textbook Campbell Biology and required powerful funding and several years to be built. We posed the question whether the process of transforming a textbook into an e-tutor can be simplified without loss of its usability. The second aim of the paper is to identify basic data model components enabling the extraction of knowledge included in e-textbooks, such as knowledge items, topics and fragments of e-textbook. The third aim is to compare our results with those achieved in the Inquire system.

The division of e-textbook into fragments stored in the database is to enable automatic generation of:
1. Questions and multiple-choice assessments verifying whether students have mastered the course material.
2. Materials supporting repetition, refreshing and retention of students’ knowledge including presentations and summaries of concepts involved.
3. Generation of different, individualized paths through e-textbook.

The proposed system is to become an interactive students’ aid assisting them in their studies. It is to help the students with commanding and retention of the course material, and it is not intended to replace the standard study process based on learning materials, homeworks, tests, projects and contacts with the instructor. It is assumed that at the beginning the domain expert will introduce e-textbook content paragraph by paragraph simultaneously identifying knowledge components such as concepts and topics. The system will store the data in the database including information about students’ performance.

The paper starts with an investigation of data models applied in such information systems as Wikimedia and Semantic Web. Next basing on those, database representation of the content of the e-textbook is built. A prototype e-tutor application operating on data from the database is presented. The application does not use base tables directly. It uses auxiliary relational views combining database content with the expressions in the natural language enabling dialogs between the system and the learner. Finally the comparison with the features of the Inquire system is presented.

References:
[1] Chaudhri V.K., Cheng B.H., et. al., (2013). Inquire Biology: A Textbook that Answers Questions. AI Magazine. 34(3), p.55-72.

Keywords:

Knowledge representation, knowledge database, e-learning automation, e-tutor.