Open learner model
A learner model, or student model, is a structured representation of a learner’s knowledge, misconceptions, and difficulties (Bull, 2004). Learner models are constructed from learner data usually gathered by an intelligent tutoring system through the learner’s interaction with the tutoring system. Learner models allow tutoring systems and instructors to understand the strengths and weakness of their students. Based on that, the instructional approach can be adapted to each specific student or group of students’ needs. An open learner model enables learners to view and analyze their learner model. It empowers learners to examine their own representation, encouraging metacognitive behaviours like self-awareness and self-regulation.
According to Kay et al. (1997), an open learner model can help students understand their learning progress and processes because it enables the student to answer questions such as:
Researchers argue that an open learner model should provide ways for the students to measure and compare their knowledge against their peers’ knowledge, experts’ knowledge benchmarks, and the knowledge they need to perform well on exams (Kay et al., 1997). According to Kay et al., 1997, in order to help the learner understand the learner model and set achievable learning goals, the model should provide a comparison standard. That is, the learner should be able to compare his or her learning level with the learning progress of peers or with a targeted level, e.g., expert knowledge or A-level knowledge. The comparisons available depend on the learners and on their goals and abilities. Based on these definitions, open learner models can be viewed as a form of cognitive tool, as they work on facilitating the learning process.
In their 2007 paper, Susan Bull and Judy Kay have proposed a framework to analyse open learner models regarding their aspects and purposes. The SMILI:) open learner model framework examines systems based on four main elements (Bull and Kay, 2007).
The open learner model presented in figure 1 reports the learner progress within the Enhanced Entity-Relationship tutor, an intelligent web tutor of conceptual database design (Thomson and Mitrovic, 2010). This open learner model showcases progress bars to represent the proportion of correct and incorrect learner knowledge. Through this interface, the learners can analyse their learning process. The system displays the portion of topics that were covered and have been mastered, as well as, a measure of incorrect knowledge, an indication that the learner should revisit those topics.
In the system discussed by Kay et al., 1997, the open learner model is displayed as a tree (Figure 2). Each of the nodes represents classes of commands to be learned in the “sam” text editor context. A full square node means that the learner has mastered that command set, while an empty square node means that the learner lacks the knowledge of that set of commands. Uncertain information about the learner’s knowledge is represented as nested squares nodes.
Learning Analytics Dashboards
Learning analytics dashboards are educational tools similar to open learner models. They also aim to enhance learning by presenting information about learners’ interaction with e-learning systems. These data-driven tools intend to encourage learner’s self-knowledge and to aid stakeholders’ (i.e., students, instructors, and administrators) decision-making processes regarding learning strategies. However, as Bodily et al., 2018, point out these dashboards are not grounded on learner modelling; that is, they do not try to create representations of learner knowledge. They do not intend to infer what the learner understands based on system interaction. Rather, they perform data analysis on interaction data for the stakeholders to use.
In their 2013 paper Verbert et al., discussed examples of learning analytics dashboards. The StepUp! tool summarises the learners' interaction with the learning system and displays it to the learners and their teachers. Figure 3 provides an overview of the tool’s learner view, which compiles the learner’s social activities within the system. As one can see, this tool does not make conclusions about learning or knowledge.
One of the primary challenges faced by open learner models is how to present comprehensible information. The learner must be able to understand what his or her representation means in order to act on it. This representation challenge is also complicated by naturally hard to visualise information (Kay et al., 1997). The system may compute information about learners that is not easily visualised or understood. The system designers have then to decide whether to hide or show this complex information. The system designers may also compute information that could potentially discourage the learners, e.g., slow progress or low retention. They are then faced with the challenge of displaying this information in a stimulating way.
Another challenge is how to establish the amount of power the learners have when dealing with their open learner model. Allowing the learners to update their knowledge representation without learning evidence may cause them to overestimate their skills, and then, hinder the system adaptability to their actual capabilities. However, preventing the learner from adding information to his or her model is not a solution. The learners may have previous knowledge about certain subjects, such knowledge is untracked by the tutoring system. If they are now allowed to add this information to their models, their experience may feel repetitive and unengaging.
- Bull, S. (2004). Supporting learning with open learner models. Planning, 29(14), 1.
- Kay, J., Halin, Z., Ottomann, T., & Razak, Z. (1997, December). Learner know thyself: Student models to give learner control and responsibility. In Proceedings of International Conference on Computers in Education (pp. 17-24).
- Bull, Susan, and Judy Kay. "Student models that invite the learner in: The SMILI:() Open learner modelling framework." International Journal of Artificial Intelligence in Education 17.2 (2007): 89-120.
- Thomson, D., & Mitrovic, A. (2010). Preliminary evaluation of a negotiable student model in a constraint-based ITS. Research and Practice in Technology Enhanced Learning, 5(01), 19-33.
- Bodily, R., Kay, J., Aleven, V., Jivet, I., Davis, D., Xhakaj, F., & Verbert, K. (2018, March). Open learner models and learning analytics dashboards: a systematic review. In Proceedings of the 8th international conference on learning analytics and knowledge (pp. 41-50). ACM.
- Verbert, K., Duval, E., Klerkx, J., Govaerts, S., & Santos, J. L. (2013). Learning analytics dashboard applications. American Behavioral Scientist, 57(10), 1500-1509.