Marc Kaltenbach

Screen management in hypertext systems with rubber sheet layouts

This paper addresses the issue of screen management in hypertext systems. It presents a new way of placing windows, or any graphical object delimited by rectangular boundaries, in the context of an existing set of windows. The heart of the technique presented is a particular re-scaling of the display which changes the locations of objects while maintaining their sizes and avoiding object overlapping. This functionality has uses both for authoring and browsing hypertext documents. In particular it enables the display of hierarchically structured information at various levels of detail and complements other visual graph management functions. More generally, the objective is to attenuate the feeling of disorientation users experience when related information obtained through hypertext browsing are stacked upon a display screen. This work suggests extending hypertext systems by enabling users to create well structured information “collages” and to program animated presentations on the basis of browsing through ill or differently structured collections of documents.

Strengthening the Novice-Expert Shift Using the Self-Explanation Effect

A critical assessment is given of a model of skill acquisition (Cascade) and of its account of the “self-explanation effect.” This model reproduces result differences observed between students with apparently the same level of intelligence and pre-requisite knowledge. However it does not seem to account for the transformation of novices into experts. We are proposing an alternative interpretation of the psychological facts, based on learning contexts, that does not have this limitation. A brief account is given of a system that would make the benefits of self-explanations more widely distributed.

LKC: Learning by Knowledge Construction

Domain knowledge is mainly communicated to learners via unassessed content such as text. Lack of effective support in reading activity of this type of resources, induces students misconceptions while increasing their cognitive load, leading to their demotivation. In this paper, we present the Learning by knowledge Construction approach (LKC). This system provides full student support for knowledge acquisition in reading activity. Aspects such as document annotation, external representations and argumentation are all taken into account. We describe an ITS architecture that implements this approach and give details on the authoring and student learning environments.

Using adaptability to create an automatically adaptive course presentation on the Web

We describe our approach to building an automatically adaptive course presentation on the Web. We aim to combine adaptability with the learner-driven course. To achieve this goal, we design and implement an agent called a confidence agent. This agent analyzes the learners behavior during a learner-driven stage and then not only adapts the next presentation but also updates the domain model accordingly.

A Conceptual and Contextual Object-Oriented Logic Programming: the PROLOG++ Language

Building upon J. Sowas Conceptual Graph (CG) theory, this paper introduces basic elements of the new language, PROLOG++, subsuming Prolog with various objet oriented, conceptual and contextual extensions. A Prolog++ “program” is composed of a declarative knowledge base and a distributed strategic knowledge base; the latter forms a network of objects that communicate by sending messages. A message corresponds to a goal described by a term or a CG (simple or compound). Declarative knowledge base corresponds to a “conceptual dictionary” describing the semantic of concepts and relations used in CG. The declarative base is composed of two hierarchies, one for concepts and the other for relations, each element of the two hierarchies corresponds to an object made up of conceptual structures. This base thus endows Prolog++ with a second form of object oriented programming. Finally, Prolog++ provides, as predefined methods, a set of conceptual operations for editing and handling CG.

Building an effective groupware system

Although, the growing number of Internet users who can play a prominent role in building collaborative groupware systems, the ability to find helpers (tutors or other learners) is still a challenging and important problem. Helpers could have a lot of useful information about courses to be taught, however, many learners fail to understand their presentations. We suggest a new filtering framework, called a pyramid collaborative filtering model, to trickle the number of helpers down to just one. The proposed pyramid has four levels. Moving from one level to another depends on three filtering techniques: domain model filtering; user model filtering; and credibility model filtering. Our experiments show that this method greatly improves filtering effectiveness.

Web course self-adaptation

This paper describes the methodology of an intelligent agent for building a self-adaptive course on the Web. An important task, therefore, is to combine adaptability with the learner-driven course in order to get a self-adaptive mechanism. For this, we have suggested new structure for a Web course. Based on this structure, we have suggested a new method to evaluate the granularity level of each segment on the course. This method evaluates the segment that a learner most prefers. To achieve this goal, we design and implement an agent called a confidence agent. Our experiment to evaluate our adaptation method shows that our approach greatly improves the domain model, and presents a course better related to the learners needs.

Reading for understanding: a framework for advanced reading support

Reading is still the main tool for knowledge communication, especially in higher education. Lack of an effective support during this activity, induces many problems of comprehension and understanding. We present the LEKC framework (learning by explicit knowledge construction), which aims to assist readers by an explicit construction of key understanding elements. It introduces knowledge models for the discourse, the epistemic tasks and the instruction.

Re-using web information for building flexible domain knowledge

Building a knowledge base for a given domain usually involves a subject matter expert (tutor) and a knowledge engineer. Our approach is to create mechanisms and tools that allow learners to build knowledge bases through a learning session on-line. The Dominant Meaning Classification System (DMCS) was designed to automatically extract, and classify segments of information (chunks). These chunks could well automate knowledge construction, instead of depending on the analysis of tutors. We use a dominant meaning space method to classify extracted chunks. Our experiment shows that this greatly improves domain knowledge.

A curriculum based visual interface for course authoring and learning

This paper proposes a visual model to help create and exploit curriculum based ITS. In the system, a curriculum is a capability transition network that consists of two kinds of nodes: capability nodes representing teaching outputs and transition nodes organizing tutoring activities. The bridge between capability nodes and transition nodes is the prerequisite links from capability nodes to transition nodes and the output links from transition nodes to capability nodes. In the model, domain knowledge is identified as human capabilities based on Gagnés instructional theory. Transition nodes are modeled after a combination of the task classifications proposed by Gagné and by Bloom. Visual properties of a capability transition network reflecting these categories as well as learner state can assist both curriculum authors and learners in their respective tasks.