Introduction
This thesis is part of research in Computer Environments for Human Learning (EIAH), aimed at stimulating, supporting, and personalizing learning experiences The field of EIAH is interdisciplinary, intersecting computer science, cognitive psychology, didactics, and educational sciences In this computer science thesis, we focus specifically on user assistance in educational software applications, conducted within the AGATE project (AGATE, 2016).
In the educational context, the use of computer applications is on the rise, with both learners and teachers being primary users The learning situations involving these applications are diverse This thesis focuses on the individual use of applications by learners.
The challenges of user adoption and utilization, along with the lack of pedagogical feedback and guidance in certain educational applications used by learners, highlight a significant need for assistance To address this need, implementing support systems and existing solutions is essential.
Les applications utilisées en contexte éducatif sont de deux types
Educational applications can include software developed through EIAH research, private companies, or directly by teachers to meet their specific needs For instance, Cabri-Géomètre, created from EIAH research, allows learners to construct geometric figures Additionally, the Maxicours platform offers courses and academic support exercises, while English OnLine France provides multiple-choice quizzes designed by educators.
In addition to pedagogical software, non-pedagogical applications, which are not specifically designed for learning, are also utilized in educational contexts For instance, computer science students learning C++ use the Visual Studio development environment in their practical sessions, where mastering this software becomes a necessary skill Furthermore, activities aimed at learning objectives that do not focus on mastering a software tool may still require non-pedagogical applications For example, in a word processing course, a teacher may have students use Microsoft Word to learn how to format official letters, while in a math class, Excel may be employed to assist with algebra problem-solving Additionally, language instructors might utilize recording software to help students improve their pronunciation In these scenarios, a lack of proficiency in the non-pedagogical application can hinder the acquisition of the intended educational knowledge Ultimately, some pedagogical activities may involve both understanding a non-pedagogical application and other subject-specific knowledge, such as when an instructor requires students to use a specific development environment like NetBeans for their assignments.
17 pratiques dans le but d’apprendre à utiliser cet environnement tout en acquérant des connaissances en programmation
Regardless of the prescribed software applications, whether educational or not, learners may abandon their use due to challenges in understanding and operating them (Cordier et al., 2010) In an educational context, the mandatory use of such software can hinder knowledge acquisition due to its complexity To mitigate this issue, providing technical support to learners is a viable solution.
Many learning applications lack pedagogical feedback and clear instructions for the learning activities they offer Choosing the right activity for a specific learner can be challenging for both students and teachers To aid in this selection, pedagogical guidance can be beneficial, yet not all applications provide it For instance, the Matou Matheux website offers a range of exercises for teachers and students but lacks guidance that would help learners find activities suited to their needs Additionally, the only feedback provided by Matou Matheux is a simple success or failure diagnosis for each exercise Educators might prefer more comprehensive feedback, such as hints to assist struggling students or examples related to the current exercise before revealing the solution.
Many non-educational applications also exhibit significant shortcomings, particularly in their lack of learning activities, as they are not designed for educational purposes For instance, tools like Excel do not offer any learning activities related to mathematics, while Visual Studio lacks programming knowledge acquisition activities These non-educational software solutions focus primarily on task completion and efficiency, aiming to simplify user experience by providing problem-solving solutions or automating actions Consequently, they fail to deliver appropriate pedagogical feedback.
Ces différentes situations ó l’on souhaite enrichir des applications informatiques existantes avec des compléments pédagogiques nécessitent une assistance pédagogique
By integrating technical support and educational assistance to enhance both pedagogical and non-pedagogical software used by learners, a comprehensive solution can be provided to overcome the existing limitations of applications utilized in educational contexts.
Mise en place de l’assistance
Teachers often serve as secondary users of the applications utilized by their students In this capacity, they typically aim to offer both technical and pedagogical support to learners Numerous examples demonstrate that educators are capable of implementing such assistance, whether or not they rely on computer applications.
When teachers ask their students to work with software applications, they provide guidance on how to use these tools effectively This support can include creating instructional videos or documents, as well as setting up the system prior to student use For instance, in a second-year Human-Computer Interaction course at the University of Lyon 1, instructors utilize Camtasia Studio to produce five tutorial videos demonstrating the use of NetBeans, a non-educational application These tutorials are incorporated into the first practical session of the Human-Computer Interaction course focused on programming with NetBeans By creating these tutorials, teachers aim to assist students in mastering the non-educational application and to clarify the learning activities they propose, which are not outlined by NetBeans itself.
Teachers often provide pedagogical feedback to struggling learners to help them overcome difficulties For instance, the Hot Potatoes application (Winke and MacGregor, 2001) allows educators to create multiple-choice questionnaires in a web format To offer effective feedback, teachers set up the tool by specifying correct answers and providing explanations for incorrect ones This feedback is then made available to students when they request to check their answers.
In terms of pedagogical guidance, teachers create tailored learning activities based on the learner's educational context Certain applications assist educators in utilizing non-pedagogical tools effectively, such as those that generate structured sequences of activities for students For instance, the EIAH OASIF (Galisson and Nouveau, 2003) supports teachers in planning learning activities Similarly, non-pedagogical applications like Word can also be employed to develop a comprehensive learning activity plan.
In addition to technical and pedagogical support through applications, the most common form of assistance in education is the direct help provided by teachers to students Teachers can assist with application usage, offer pedagogical feedback, and sometimes guide students to more relevant activities However, this method is not optimal as it requires the constant presence of the teacher, leading to frequent issues of teacher overburdening To aid teachers in monitoring and supporting students, numerous studies on tutoring have developed supportive tools for educators.
19 outils peuvent aller jusqu’à proposer une assistance informatique automatique apportée aux apprenants et permettre donc de décharger l’enseignant de ses interventions (Labat, Pernin and Guéraud, 2006)
To address the support needs of learners faced by educators, an early solution in the application lifecycle is to modify the source code However, this source code is not always accessible, and even when it is, teachers—who may lack programming skills—find it challenging to make modifications In such educational contexts, it is essential to implement assistance that does not require altering the source code or possessing programming expertise Therefore, the establishment of epiphytic assistance systems is a suitable solution An epiphytic assistance system is designed for users of a specific application and can be integrated into that application without disrupting its functionality (Paquette et al., 1996).
État de l’art sur les systèmes d’assistance en contexte éducatif
Les types d’assistance
Technical assistance is offered to alleviate challenges in understanding and using applications In an educational context, these challenges can hinder learners' knowledge acquisition A bibliographic study by Ginon highlights the importance of addressing these difficulties to enhance the learning experience.
2014) a identifié, catégorisé et hiérarchisé les besoins d’assistance technique (cf Figure 2) Nous repartons de ce travail récent et complet pour le compléter par notre classification des assistances en contexte éducatif
Figure 2 : Classification des caractéristiques de l’assistance technique (Ginon, 2014)
(Ginon, 2014) a identifié trois principaux besoins d’assistance (cf T dans la Figure 2) : la découverte du système (T.1), la réalisation d’une fonctionnalité (T.2) et l’amélioration de la pratique (T.3)
Assistance in discovering the system can enhance understanding of its functionalities and features Users need to grasp the purpose of the system, the terminology and objects involved, how to utilize them, the offered functionalities, and the necessary steps to complete tasks effectively.
To assist with task completion, support can offer feature discovery, task tracking, and potentially task facilitation Tracking progress can help users better understand their actions, especially during lengthy and complex tasks.
The process of enhancing user understanding can be achieved through projection, validation, and assessment Projection involves showcasing the user's completed work to facilitate comprehension and identify areas for improvement, often through previews or examples Validation confirms the correctness of the user's efforts or highlights any inconsistencies An assessment summarizes the user's progress, clarifying what has been accomplished and what remains to be done Additionally, task execution may create a need for guidance to assist the user in determining their next steps Finally, user tasks can be simplified by partially or fully pre-completing them, allowing for modifications or additions to the suggested work.
To enhance user performance in task execution, assistance can suggest enriching the user's practice by highlighting unused options or features provided by the system Additionally, it can save time in task completion through the automation of certain task components.
The identified assistance needs span various domains and applications, catering to a diverse range of users from novices to experts Technical support for system exploration, task execution, and practice enhancement promotes effective engagement and utilization of computer systems Consequently, exploring the system facilitates user onboarding, while assistance with task completion and practice improvement encourages ongoing use.
Pedagogical assistance is central to an educational support system Unlike technical assistance, which aims to help learners complete tasks quickly and without errors, the primary goal of pedagogical assistance is to enhance the acquisition of knowledge through engagement.
In the educational context, learners gain knowledge through engaging in learning activities Educational assistance should enhance or complement these activities within targeted applications Furthermore, for each activity, this assistance can provide pedagogical feedback to support the learner's execution of the task For a series of activities, educational assistance can offer guidance to help learners choose which learning activities to undertake.
In our context, a learning activity refers to the tasks that learners must complete within the application to achieve specified educational objectives, such as acquiring knowledge or skills, assessment, and remediation.
This study focuses on activities linked to digital resources such as web pages, software windows, exercises, questions, and files utilized by learners While activities can also involve non-digital resources like paper handouts and calculators, our research specifically examines those that employ digital tools Furthermore, we concentrate on learning activities defined within target applications, which may be either pre-existing or newly created.
Most educational software offers learning activities for students, such as ActiveMath, which provides courses and exercises in mathematics (Melis et al., 2001) However, some software, particularly non-educational programs like NetBeans, Word, and Excel, do not include any learning activities.
To enhance existing applications, teachers can incorporate learning activities that were not originally included Several platforms, such as ASKER, MIRTO, and Exercise Maker, enable the generation of exercises For instance, ASKER allows educators to create exercises through an authoring interface for self-assessment by learners However, many applications, especially non-educational ones like NetBeans, do not support the creation of such learning activities In these cases, teachers can propose external learning activities by providing videos or documents that explain these activities.
Figure 3 : Instructions explicitant l’activité dans AMBRE-add
(Nogry, Guin and Jean-Daubias, 2008)
Clearly outlining the learning activities within the application is essential for learners to understand their tasks Instructions can be delivered in various formats, including target applications, printed or digital documents, and videos For instance, AMBRE-add (Nogry, Guin, and Jean-Daubias, 2008) provides messages to clarify activities, such as "Choose a diagram to represent the problem" or "Show what we are looking for," guiding learners through each step Additionally, the software Tri sélectif (Jean-Daubias, 2011) offers similar support to learners.
32 une activité sous forme d’un jeu dans lequel ils doivent mettre les déchets dans la poubelle pertinente Des instructions explicitant cette activité sont prononcées oralement par le logiciel
To engage in learning activities offered by applications, learners utilize various digital resources such as web links, application content, and files Teachers or applications can assist learners in accessing these resources by providing instructions or directing them to the relevant materials For instance, Aplusix organizes its exercise resources through a structured exercise map that categorizes tasks based on the type of numbers involved (integers, decimals, fractions, radicals) and their difficulty levels, ranging from 1 to 10 Similarly, in the case of ASKER, teachers can share links with students to facilitate access to specific exercises within the application.
Learning activities are essential in educational contexts However, the desired learning activity may not always be available in the application being used An assistance system can provide a solution by allowing teachers to add or enhance learning activities, while also guiding students in using digital resources or clarifying the tasks to be completed.
Figure 4 : Carte d’exercices sur Aplusix
Les modes de déroulement d’une assistance
The term "assistance flow" refers to the interaction between various elements that make up an assistance system, such as messages or highlights These elements can be sequenced in different ways; for instance, help messages may be displayed one after another or simultaneously Our research has revealed patterns in the flow of the assistances we studied In this section, we illustrate, with examples beyond the educational context, the different modes of interaction between the elements of assistance that our study has identified.
In certain applications, assistance elements are provided independently, with no dependencies between them Each assistance element is triggered by its own specific events For instance, a help tooltip appears when the user hovers over a particular component of the application This method of interaction is referred to as independent mode.
Figure 19 : Exemple de l’affichage de la vue d’ensemble d’un cours d’IXL Learning (IXL, 2016)
IXL Learning presents various courses to learners through independent messages displayed in pop-up windows When a learner hovers over a course, a description appears, such as the "Counting review - up to 10" course, which is illustrated with a sample question In contrast, ActiveMath provides explanations of mathematical concepts through pop-ups or web pages.
In certain applications, assistance is provided in a step-by-step manner, where one element follows another This sequential guidance is commonly found in tutorials that help users learn how to use an application effectively We refer to this method of articulation as the successive mode.
Figure 20 : Exemple d’une étape d’assistance dans le tutoriel intégré à Connectify (Connectify,
2016) et d’une étape de vérification des réponses dans EOLF (EOLF, 2016)
The Connectify application allows users to create a hotspot for sharing internet access It provides sequential messages to guide users in utilizing the app effectively Additionally, these messages can be used to explain user errors step-by-step.
Web forms often allow for simultaneous display of user input diagnostics, providing a comprehensive view of the information This enables users to easily identify the assistance they require We refer to this method of presentation as simultaneous mode.
Dans le contexte éducation, nous avons, par exemple, Hot Potatoes (Winke and MacGregor,
In 2001, a tool was developed that enables educators to create various types of exercises in the form of web pages, providing diagnostic feedback on students' responses This diagnostic feature allows for simultaneous evaluation of multiple answers For instance, an exercise created using EOLF with Hot Potatoes at the University of Franche-Comté displays indications of "correct" or "incorrect" for all responses at once.
Assistance can be provided progressively, adapting to the frequency with which a user encounters the same problem Specifically, the level of assistance becomes increasingly important—more detailed and precise—when addressing recurring needs This approach is referred to as a progressive mode of support.
In ActiveMath (Melis et al., 2001), learners can seek progressively detailed assistance when solving exercises Initially, help is provided as a hint (1), followed by a more detailed clue (2) upon a second request for assistance If the learner still struggles, a complete solution (3) is offered on the third request to help them overcome obstacles This approach ensures that the support becomes increasingly specific and concrete as needed.
Some applications leverage various information sources, such as application status and user profiles, to deliver more relevant assistance This results in personalized and contextualized support, enhancing the user experience We refer to this approach as an interactive mode.
Figure 21 : Exemples de vérification de la réponse sur ActiveMath (Melis et al., 2001)
ActiveMath (Melis et al., 2001) and Hot Potatoes (Winke and MacGregor, 2001) evaluate the application's state by analyzing the text written by the learner ActiveMath checks the learner's responses and displays the results, highlighting three different diagnostics: syntax error, incorrect answer, and correct answer This diagnostic process relies on the consultation of the learner's response.
2.2.6 Classification des modes de déroulement d’une assistance
This section of the literature review explores the various modes of assistance in educational contexts, leading to a proposed classification illustrated in Figure 22 We have identified five distinct modes of assistance: independent, sequential, simultaneous, progressive, and interactive.
Figure 22 : Classification des modes de déroulement d’une assistance
Mise en place d’assistance en contexte éducatif
This section focuses on the implementation of assistance in educational contexts Assistance can be provided by advisory systems, human or artificial tutors, within a pedagogical scenario, or integrated into adaptive hypermedia Additionally, assistance may also be a feature of the application itself.
The process of assistance is shaped by various tasks that also outline its execution methods However, these execution modes often remain implicit, as they are not central to the tasks at hand To complement this understanding, we have examined Grafcet, which emphasizes the progression of a system, even when that system is not specifically an assistance system.
We compared our approach to implementing assistance—characterized as epiphytic, generic, and requiring no programming skills—with the identified features of educational assistance Each system studied is presented alongside a table that positions it according to the following characteristics: no programming skills required, epiphytic approach, generic approach, technical assistance, pedagogical feedback, pedagogical guidance, and the progression of assistance The results of this comparison are visually represented across three levels: permitted (✓), permitted but limited (✓), and not permitted (✗).
A recommendation system is designed to actively assist users of specific software by analyzing their actions and outputs Notably, the generic recommendation system Marco and the Epitalk system stand out due to their generic and epiphytic approaches.
The Epitalk system (Paquette, Pachet, and Giroux, 1994) enables the implementation of epiphytic advisory systems for target applications developed in Smalltalk-80 through a generic approach This advisory system generates recommendations by observing the interactions between a user and the target application.
The architecture of the advisory system in Epitalk is based on a task graph that decomposes user actions within target applications into multiple hierarchical levels down to terminal tasks This graph helps identify the user's current task, enabling the provision of relevant advice The assistance designer specifies the advice along with its associated metadata, including an importance level that guides the system in selecting the most pertinent advice when multiple options are available Additionally, advice can be linked to a maximum number of uses before temporary deactivation, and the designer can decide whether the advice is triggered by the end user or initiated by the advisory system.
Epiphytic agents display advice organized within a graph structure that is automatically generated and isomorphic to the task graph Each task is associated with a specific agent The primary functions of an epiphytic agent include receiving events from the target application sent by spies and providing advice by executing rules in accordance with the NéOpus formalism.
Figure 23 : Édition d’un graphe de tâches dans Epitalk (Paquette, Pachet and Giroux, 1994)
La Figure 24 confronte les spécificités du système conseiller Epitalk à notre approche (épiphyte, générique, pas de compétences de programmation) et aux caractéristiques de l’assistance en contexte éducatif
Epitalk employs an epiphytic approach tailored specifically for Smalltalk-80 applications, requiring designers to have knowledge of Smalltalk-80 programming While Epitalk minimizes the need for extensive programming, it remains challenging for non-computer science educators to access and utilize effectively.
Epitalk enables the specification and execution of assistance while providing pedagogical feedback on target applications in a streamlined manner, offering users helpful advice Educators can utilize it to deliver focused pedagogical guidance on Smalltalk-80 applications by defining recommendations regarding activity selection and the criteria for transitioning between activities within the rules associated with each agent.
Figure 24 : Epitalk confronté à notre approche et aux caractéristiques de l’assistance en contexte éducatif
Richard (2008) proposes an approach for implementing advisory systems on existing web pages using an epiphytic method These advisory systems aim to provide users with navigation guidance within target pages of a website, triggered by user clicks on links The advice is delivered either through advice pages embedded on the target pages by the advisory system or through new pages created or transformed from the target pages The former appears as a pop-up window on the target pages, while the latter opens as independent pages.
This approach utilizes a task model, which is a subset of the website graph comprising nodes that represent links and arcs that indicate transitions between these links This model effectively simulates user navigation actions, specifically the clicks on links To trigger advice during these clicks, a usage model must be developed, linking the advice to the identified arcs within the task model.
Based on this approach, a generic advisory system architecture called Marco has been proposed To implement an advisory system on a website, it is essential to instantiate it with the site's task model, the advice provided, and the usage model.
This approach has been validated through Marco's implementation on two sites: EducaSource and Apache Figure 25 illustrates an example of advice presented as a pop-up window on a web page of the EducaSource site (EducaSource, 2017).
Figure 25 : Exemple d’un conseil sous forme d’une fenêtre pour EducaSource avec Marco
La Figure 26 confronte les spécificités du système conseiller générique Marco à notre approche (épiphyte, générique, pas de compétences de programmation) et aux caractéristiques de l’assistance en contexte éducatif
The Marco generic advisory system is an epiphytic and versatile solution designed to be instantiated for creating specific advisory systems tailored to various existing web applications However, it is not entirely generic, as it is specifically focused on web-based applications Furthermore, this system is dedicated to assisting users during their navigation on a website, facilitating the transition from one page to another through link clicks.
It is not feasible to provide advice based on various interactions on web pages Marco offers technical support, specifically in system exploration and practice enhancement, but this assistance is limited and cannot automate actions on web pages While teachers can utilize it to create support, its capabilities are restricted If an activity is linked to a target site page, teachers can develop guidance on navigating web pages and provide feedback for each click on a link.
Figure 26 : Marco confronté à notre approche et aux caractéristiques de l’assistance en contexte éducatif
Discussion
Dans ce chapitre, nous avons classifié les caractéristiques de l’assistance en contexte éducatif notamment (cf Figure 43) :
• l’assistance technique pour la prise en main et l’utilisation de l’application, en exploitant la classification de l’assistance technique proposée par (Ginon, 2014) ;
• l’assistance pédagogique qui permet de proposer des feedbacks pédagogiques et un guidage pédagogique avec 5 types de guidage pédagogique : guidage libre, guidage temporel, guidage contextualisé, guidage personnalisé, guidage séquentiel ;
• le déroulement d’une assistance avec 5 modes de déroulement : indépendant, successif, simultané, progressif, interactif
We examined existing research that offers assistance or facilitates the implementation of identified aids in educational contexts Our analysis compared these works to our classification of assistance and our approach to implementing support that does not require programming skills, adopting an epiphytic and generic approach Figure 44 highlighted the limitations of these studies in light of this comparison We found that most of the works do not adopt a generic approach; they are specific to an application, environment, or domain For instance, the Marco system is generic compared to specific advisory systems that instantiate it, but it only assists with navigation between web pages Conversely, some studies target specific goals but can be adapted for other purposes; for example, the Lightweight Tutoring Agents model aims to create an intelligent tutor that primarily provides pedagogical feedback, yet it can also be utilized for technical assistance Lastly, to our knowledge, no studies have explored the modes of assistance delivery, which, while present in some systems, are not explicitly represented.
Figure 43 : Classification complète des caractéristiques d’une assistance en contexte éducatif
Figure 44 : Les travaux connexes confrontés à notre approche et aux caractéristiques de l’assistance en contexte éducatif
En conclusion, aucun de ces travaux ne propose de solution complète pour mettre en place un système d’assistance sur des applications existantes variées en contexte éducatif
As part of the AGATE project, a generic approach was adopted to propose models and tools for implementing epiphytic assistance in target applications without requiring modifications to their source code or programming knowledge These efforts were designed to provide technical assistance In the following chapter, we will present the AGATE project, its contributions, and explore the potential to leverage them for adding technical and pedagogical support to software in an educational context.