Some characteristics of Instructional Design. for Industrial Training

Some characteristics of Instructional Design for Industrial Training Claude Frasson Université de Montréal, Département d'informatique et de recherche opérationnelle 2920 Chemin de la Tour, Montréal, H3T 1J4, Québec, Canada E-mail: frasson@iro.umontreal.ca Abstract. Industrial training needs to be improved as employees are faced with a rapid change in their knowledge environment and their responsibilities. Instructional design was often considered as a keypoint of knowledge transfer but the outcomes in terms of cost and efficiency depend greatly on how the design is realized and how the employees can efficiently use the learning material. In the SAFARI project, which aims at developing various components of Intelligent Tutoring Systems (ITS), cooperation with industry led us to deliver gradual tutoring systems that correspond to different real needs. Lessons learned from this experience highlight some realistic aspects of training in industry and allow to consider instructional design according to a new point of view. 1. Introduction Industrial training is of major importance for economic competitiveness. Employees have to face two important waves of changes in their environment : the rapidity of penetration of advanced technologies and the transformation of their responsibilities. They have to acquire both new knowledge on more complex environments and multiple skills, and learn new roles. However, industrial training is time consuming and expensive. For instance, IBM US spends $2 billion a year on training including $1 billion for trainers salaries; each year, the US government spends $20 billion on military training [1]. To this costs one should add the cost of stopping production tasks, pulling the employee out of the workplace during training sessions. Updating existing courses is also long and expensive as very few real database systems are used to store and maintain the courses. Presently, training in enterprise consists essentially in giving a course using transparencies or an LCD. There is no real evaluation of the learner to know how the course is finally mastered and well integrated to his previous knowledge. This superficial training needs to be completed by several additional weeks or months of practice in real situation. There is no consideration for the characteristics of the learner, learning style, conceptions and misconceptions. In particular, it is very

difficult to detect misconceptions acquired by the employee during the training and especially to eliminate them, except when problematic situations involving costly errors have occurred. During the last years, a major misconception was to consider that industrial training should be essentially based on a more or less participation of employees to a course for which instructional design had to be defined. The needs of industry are different and attending a course is only one way of training that should be adapted to the employee. A dynamic adaptation to the learner requires a deep knowledge of his behavior in learning situations and more particularly during the resolution of problems. This implies to recreate learning situations close to the reality in which employees will be placed. Cooperation with industry through the SAFARI 1 project led us to consider a new approach for instructional design. In this paper we will present the main issues of this project and the impact on a new methodology of training involving gradual steps. 2. The needs of Industry The Synergie program aims (1) to enhance cooperative research and development between universities and industry, (2) to accelerate the product development cycle, (3) to facilitate the transfer of knowledge between research establishments and the industry, and (4) to educate highly qualified professionals in the domain. The main objectives of SAFARI [2] are to develop a methodology and an environment for the creation of tutoring systems to be used in professional formation. The focus is on teaching mostly procedural knowledge concerning the operation of devices such as medical instrumentation, consumer appliances and aeronautical instruments. SAFARI involves four Québec universities, two private enterprises and a government agency. The industrial partners are Virtual Prototypes Inc., providing a simulation software package VAPS, and Novasys Inc. VAPS (Virtual Applications Prototyping System) is a high-quality commercial interface-building and simulation system, used in many areas (such as airline cockpit design). Our findings when evaluating the industrial training needs were the following: in new tasks to achieve, employees prefer to be assisted by a kind of adviser able to supervise or advise the progression of work to be realized, employees have practically very few time to spend on a course; frequent situations show short term needs to receive an answer to their preoccupation, 1 SAFARI is a project under the auspices of Synergie, a programme sponsored by the Ministry of Industry, Trade, Science and Technology of the Government of Québec

a course should be practically adapted to each employee taking into account his previous experience and personal characteristics, a majority of situations should benefit from a pedagogical approach based on learning by doing. We observed that a natural cycle in which most people acquire a given skill is by first observing someone s demonstration of the skill, then freely experimenting with the device in question (given the availability of the device, and that such experimenting is not hazardous),then executing precise tasks (assignments) in terms of the device functionalities under the guidance of an expert, and finally by communicating the learned skill to another person. So we developed various prototypes of training systems able to provide the following learning modes: demonstration: the system presents a simulation of various tasks to the learner. The realization of the tasks depends totally on the system without intervention of the learner (for example, a situation where the learner observes the computer solving a problem), free exploration: the learner can navigate into a simulation system which reacts to his actions without intervention or guidance of the system. The learner controls his activity (for example, the resolution of a task) and this mode can be compared with free navigation within a hypermedia document, advice: the learner is in a problem solving situation and can benefit from advice of an adviser [3] who continually watches the tasks and can correct the actions with in depth explanations. Various types of guidance (on demand, automatic, with multiple explanations,...) can be obtained, curriculum: the learner enters in a learning session, through a complete course with problems, exercises and evaluation of different activities. The course is given using a variety of learning strategies that can be selected according to the learner s model. In the first three situations instructional design will be restricted to editing problems solving situations. The last mode has two important characteristics 1. The course derives from a curriculum [4] which is organized according to a network of knowledge transitions based on capabilities (according to Gagne [5]), instructional objectives and pedagogical resources (learning materials). Achievement of instructional objectives contributes to the acquisition of capabilities. Three knowledge structures implement domain, pedagogical and didactic aspects of a subject matter through a network organization of capabilities, of instructional objectives defined on these capabilities and of pedagogical resources supporting the completion of instructional objectives. 2. Several learning strategies are provided to improve learner knowledge acquisition. According to the characteristics of the learner [6] (learning style, knowledge level) which result from the interaction with the tutoring system, a learning strategy is selected. These strategies can be carried out by different

tutoring agents [7] called actors 2 that can play different roles according to the conditions of learning. These conditions are determined by the learner s actions which are analyzed and stored in the learner s model. An architecture based on actors has been determined in [7]. For instance, we distinguish the following pedagogical actors: a tutor who gives a course according to a prescriptive approach, a co-learner who is a simulated learner with approximately the same level of knowledge than the learner, a companion, a simulated learner [8]who can give advice, the inverted tutor [9] played by the tutor who is waiting for explanations from the learner, the troublemaker [10], a particular companion who can randomly give true of false advice. Unlike the companion, the knowledge of the troublemaker is quite superior to that of the real student. A first experiment (learning of highway code) has shown that the troublemaker becomes efficient for advanced students [11]. The originality of our approach in building a generic ITS is to allow the use of multiple learning strategies. The variety of learning modes provides employees with the facility to look at a demonstration, learn using a simulation or a direct manipulation of the environment, or attend a complete course, different means that depend of their availability and objectives. 3. Generating a course from a curriculum In SAFARI the curriculum is a structured representation of the subject matter in terms of capabilities, objectives and resources. Capabilities are connected using semantic links of analogy, generalization, abstraction, aggregation and deviation. Objectives are linked using three kinds of relations: prerequisite, pretext (if an objective can contribute to the achievement of another objective), and component (if an objective contains several sub-objectives). Resources represent the way to support learning ( exercise, problem, simulation, demonstration, document,...). A course [4] is a structured set containing three categories of objectives: global, specific and terminal objectives. A global objective is a statement expressed by the teacher to globally describe all the changes (cognitive, affective) he wishes to induce in the students behavior during a course; a specific objective describes a set of behaviors that the learner should be able to demonstrate (capabilities the learner should acquire); a terminal (or operational) objective is, in our context, an objective which describes a precise performance the student should achieve. In general, a specific objective is composed of several terminal objectives. The most important 2 In SAFARI, an actor is an intelligent agent that has the capability to learn from new situations

part of a course is the course graph which is extracted from the curriculum and a flat organization of the objectives connected by appropriate links. The organization of the curriculum [4] into three networks with predetermined links, as indicated above, allows a generic approach that transforms the role of the designer. He now has only to specify the knowledge he wants to teach by selecting a set of objectives, supervising the process and approving or not the generated course. The transition networks allow to generate a corresponding course. A set of tools have been defined to allow rapid modifications of objectives or resources, allowing the designer to concentrate on specific aspects such like the course structure, the pedagogical resources, the tutorial strategies, the definition of various themes and subthemes with associated objectives, the relevance of the course or some of its objectives, etc...he also can evaluate the design using a simulation of the course. In SAFARI this approach is strengthened by an automatic course generation process that reduces considerably the amount of time devoted to the design of a course. To generate a course from one or several curricula built according to the networks indicated above, we consider the following parameters: a final state corresponding to the goals of the course to be generated. This state can be expressed either as a set of objectives that the course should reach, or as set of capabilities to be acquired by the student. We use the term Knowledge To be Transferred (KTT) to designate this set. a starting point corresponding in reality to the knowledge state of the target group. We represent this state as a vector whose elements are couples (capacity, level of mastery). We think that a good cognitive analysis of a domain can allow to identify and to classify categories of students with a more precise idea on their knowledge. The process of course generation [12] consists in considering these two parameters, and use the structures of the curricula to lead to a reasoning that will allow to define a course which is relevant for the concerned public (figure 1).

Editors Curriculum networks KTT specification Rules Base -graph generation algorithm Target group Target group -graph graph structuration algorithm structuration interface database Fig. 1: Process of course generation A target group is a group of learners with a state of knowledge on various capabilities which can be part of several subject matters (curriculum). The course graph is generated by attributing to each prerequisite link a value within {acquired, partially acquired, not acquired} and to each capability a value within {possessed, partially possessed, not possessed} according to the considered target group. Rules have been determined to realize this type of marking. Having a course graph some heuristics have been developed in order to deduce, starting from each capability to acquire the objectives that are necessary to achieve these capabilities. Three type of heuristic have been implemented and have shown satisfactory automatic courses generations. 4. Generating a course from practice in the workplace One of the most important point to consider when generating a course is to be sure to have a reliable learner model. We think that a new approach in instructional

design is to try to extract the elements of the target public while the learner is performing a task in the workplace. This corresponds also to an important constraint mentioned above: the need for the employees to receive a training on-the-job due to the reluctance of management to let employees on leave for training. The advantage of this approach [figure 2) is also to observe the learner while he/she is performing a task, for instance in a simulation-based environment or on a real task, and to mark all the knowledge elements on which he/she seems to have some difficulties. Simulation Curriculum Interactive demonstration Learner model On the job training Target Group Fig. 2: On site course generation The organization of a curriculum and the attached heuristics allow to deduce all the knowledge elements that need to be acquired. The result is a potential entry for the target group which is at the basis of the course automatic generation process.this kind of approach is called in SAFARI "strategic adviser". The system observes the learner in problem solving situations and detects knowledge that should be improved or on which a training session should be created. The approach presents the advantage to be adapted to the learner s behavior in real situations. 4. Conclusion Industrial training is of high importance for economical reasons but the needs of enterprises for training systems depend on various parameters such like the size of the company, the necessity to be at the fine point of technology, the domain,... In the majority of cases a simulation-based training environment will be efficient and

consequently tools for providing interactive demonstrations or simulations will be greatly seeked by the industry. In that case instructional design will be concentrated on the decomposition of a space of problem-solving situations. However when training requiresa curriculum and complete courses have to be built it is important to reduce the production costs and the time to deliver and maintain the learning material. Solutions adopted in the SAFARI project provide the advantage to encompass a large variety of these two kinds of situations. Instructional design tools need to be efficient and easy to manipulate in order to be effectively used by designers. The curriculum approach in SAFARI provides an automatic course generation once a curriculum is built. An additional feature is to allow direct construction of a course adapted to the learner while he is involved in problem solving situations. This new approach in instructional design should be emphasized in the next years. Acknowledgments This work has been supported by the Ministry of Industry, Trade, Science, and Technology (MICST) under the Synergy program of the Government of Québec. The project involves several persons who have contributed to various parts. Among them Gilles Gauthier and Roger Nkambou are at the basis of fundamental elements of the Curriculum. References 1. Murray, T., and Woolf, B. Design and implementation of an intelligent multimedia tutor. In AAAI 93 tutorials. 2. Gecsei, J., Frasson, C., SAFARI: an Environment for Creating Tutoring Systems in Industrial Training. EdMedia, World Conference on Educational Multimedia and Hypermedia, Vancouver (1994). 3. Jonassen, D. H., Grabinger, S., Instructional design and development advisor: Intelligent job aid, help system, and intelligent authoring system. Denver, CO: Division of Instructional Technology, University of Colorado, Denver (1991). 4. Nkambou, R., Gauthier, G., & Frasson, C., An authoring environmenmt for curriculum and courses building in an ITS. In 3th International Conference on Computer Aided Instruction in Sciences and Engineering. Springer-Verlag, Berlin (1996). 5. Gagné R.M., (1984).The conditions of learning, 4 ed, Les éditions HRW Ltée, Montréal. 5. Nkambou, R., Lefebvre, B., Gauthier, G.: A Curriculum-Based Student Model for Intelligent Tutoring System. Fifth International Conference on User Modelling, Kailua-Kona (1996).

7. Frasson C., Mengelle T., Aïmeur E., Gouardères G.: An Actor-Based Architecture for Intelligent Tutoring Systems. Third International Conference ITS 96, Montréal. Canada, LNCS (1996). 8. Chan, T.W., Baskin, A.B.: Learning Companion Systems. In C. Frasson & G. Gauthier (Eds.), Intelligent Tutoring Systems: At the Crossroads of Artificial Intelligence and Education, Chapter 1, New Jersey, Ablex Publishing Corporation (1990) 9. Palthepu, S., Greer, J., McCalla, G.: Learning by Teaching. The Proceedings of the International Conference on the Learning Sciences, AACE (1991). 10. Aïmeur, E., Frasson, C. & Alexe, C. Towards New Learning Strategies In Intelligent Tutoring Systems. Brazilian Conference of Artificial Intelligence SBIA'95, Springer Verlag, (1995). 11. Aïmeur, E., Frasson, C.: Analyzing a new learning strategy according to different knowledge levels, Computer and Education, An International Journal, to appear (1996). 12. Nkambou, R., Frasson, M. C., & Frasson, C., Generating s in an Intelligent Tutoring System. In proceedings of IEA-AIE 96, (1996).