This page presents a set of interactive diagrams that describe how CO3PEDIA works, from the user journey to the conversational agent architecture and the pedagogical scripting form.
Figure 1. Overview of CO3PEDIA’s main states and the general transitions between the logged-out menu, logged-in menu, scenario creation, and the management of scenarios/documents.
Figure 2. User flow in the logged-out menu: access to public pages, language switching, and authentication paths (login/registration) leading to the logged-in menu (Figure 3).
Figure 3. Logged-in menu of CO3PEDIA, including access to the dashboard (profile, presentation, management of personalised documents) and to the flows for creating scenarios (Figure 4) and managing scenarios (Figure 5).
Figure 4. Scenario creation and editing flow: completing the form, requesting AI suggestions, then choosing between local draft, saving or publishing, with a return to the logged-in menu (Figure 3).
Figure 5. Management of public scenarios (viewing/copying) and private scenarios (restricted access, resuming, deleting, downloading), including personalised documents originating from the dashboard (Figure 3).
Figure 6. Functional journey of a registered user in the Scenarios application: consulting their scenarios, duplicating/deleting, resuming, saving and publishing, with access to AI/RAG suggestions, the scenario database, and both default and personalised vectorised knowledge bases (up to 15 files).
Figure 7. Functional journey of an unregistered visitor: changing language, filling in a scenario with AI suggestions based on the default vectorised knowledge base, and consulting the list of public scenarios stored in the database.
Figure 8. Processing pipeline of CO3PEDIA’s RAG conversational agent when the user clicks the “Suggest a response with AI” button or interacts directly with the Coco chatbot. The initial request (implicit question from the form or free-text message) is combined with Coco’s role (a system prompt specialised in instructional design), the content of the pedagogical scripting form, the conversation history, and relevant segments retrieved by the RAG from a vectorised knowledge base (by default, and for logged-in users, personalised with their own documents). Together these elements form a complete prompt sent to the language model, which produces an augmented response returned to the interface and reinjected into the history. Adapted from Miladi, F., Psyché, V., & Lemire, D. (2024). “Leveraging GPT-4 for accuracy in education: A comparative study on retrieval-augmented generation in MOOCs,” AIED 2024.
Figure 9. Detailed scenario-creation journey in CO3PEDIA: starting from section 1 (Collaboration), the user successively fills in context constraints, preliminary information, technical and pedagogical information, then scenario writing (activities), productions, assessments and the schedule. Each section offers AI help buttons (see Figure 8 for the RAG pipeline). Once the form is completed, the user submits the scenario and chooses either to make it public (viewable by all) or keep it private (visible only to the author).
This part restates, in textual form, the definitions and guidance associated with the different sections of the scripting form, for more detailed consultation alongside the diagrams.
A collaborative project involves bringing together teachers and techno-pedagogues from different contexts. It makes it possible to plan collaborative learning for learners.
Collaboration makes it possible to address issues specific to students or teaching methods collectively, benefiting from different perspectives.
The collaboration agreement defines everyone's roles and formalizes them. At this stage, it may be an informal agreement.
Didactic contextualization is the active consideration of contexts in didactic practices (Blanchet, 2009). This approach emphasizes the fundamental role played by learning situations in cognitive development. Context here refers to the interactive dimension of learning, where we are led to take into account the environment in which learning takes place (Anjou, C. (2018) According to Gilbert (2006), teaching based on context effects takes on four dimensions
1- a social, spatial and temporal framework in which mental encounters with concepts are situated
2- a learning environment for these encounters specifying how actors' tasks and events are constructed and used to bring about the notions
3- the use of a specific language related to these notions
4- and a relationship with extra-situational knowledge. (Forissier, T., Bourdeau, J., & Psyché, V. (2017)
Indicate the list of authors of the collaborative scenario in context. The number of authors is for information only and is not linked to the number of contexts. For example, two teachers may work for one class (1st context), while one teacher may be assigned to another class (2nd context).
That's why the collaboration project has been set up.
Setting up a learning project based on the Context Effects model requires exchanges and collaboration between the players in the different contexts (instructional designers, teachers, learners). These exchanges take place over the Internet. It is therefore necessary to find out about Internet access and the quality of the network and bandwidth in the different areas where the project will take place. The Internet can also be used for other purposes, such as information retrieval.
This sub-section must be filled in collaboratively based on the contexts involved, so the spaces will be the same on both sides: check the virtual space(s) with which learners will collaborate and carry out synchronous and asynchronous exchanges. It is also important to define the virtual spaces in which teachers will collaborate and exchange
In this section, you need to specify the hardware required for exchanges between different contexts. It is important to check that the equipment is compatible on both sides.
Examples of Material equipment needed for exchanges:
The subject of the learning session common to all contexts
How will the project be named? Keep the title short but to the point.
The main subject of the study, it will be possible to assign specific subjects to each team later in the process.
Describe the project problem. Contextualize the specific object of study.
Enter the language(s) spoken by the learners. If more than one, list them in order of preference, from most to least used.
Indicate the constraints to be taken into account when developing content and activities specific to each context.
The pedagogical model implemented here includes original activities to support learning. These activities can be carried out within the framework of investigation in context, and include exploratory activities such as field trips, documentary surveys, interviews with players in the field, laboratory experiments, visits to companies, facilities, museums, or any other location outside the school. These activities can also be carried out in the context of a production by the learners (presentation or talk, creation of models or mock-ups, activities on specialized software or artistic activities, for example)
This section allows you to anticipate the productions that learners will have to carry out. These productions can be carried out individually or in groups.
This section allows you to anticipate what learners will be assessed on. Learners can be assessed on skills, throughout the project or during specific sessions, they can be assessed on productions, or during exams (questionnaires, exercises). Evaluators can be a variety of people (teachers, learners in the context of self-assessments, or external contributors, for example). If you wish to assess the learner on a production or skill, these can be selected from the list of previously implemented productions or skills, otherwise you can create a new assessment.