CLC - Cognitive Learning Companion - IBM Research Africa and India - overview
The education industry is witnessing rapid adoption of digital learning content and personal devices such as mobiles/tablets. This opens the door for anytime, anywhere learning. Blended learning, which extends traditional classroom learning with computer mediated learning in and out of class, is increasingly being adopted by education systems world-wide. However, these dual approaches to learning - involving human educators on the one hand and online systems on the other -- are often not well integrated (at best), or are disconnected. This leads to a fragmentation of the overall learning and teaching experience. For example, there may be limited or otherwise no information sharing between traditional classroom learning (delivered by a human teacher), and outside class learning using an automated tutor, which mainly focuses on exam-centric tutoring. As a result, existing online systems are often perceived as an additional burden that lacks context, rather than a natural partner that provides real value-add and actionable insights to the teacher. At times, issues of control over the teaching process may cause computer mediated learning to be even perceived as a threat (“will human teachers be replaced?”).
Blended learning is further challenged in resource constrained regions (e.g., Africa, India) with a different and unique set of problems such as high pupil/teacher ratio, intermittent connectivity, several contextual/environmental factors that impact learning etc. These further strain the tenuous inside and outside class connections. For learning to be truly blended, there has to be a seamless integration between in-class and outside class experience – both for the student and teacher – while keeping the parents engaged with the learning process.
To address this, we are developing a Cognitive Learning Companion (CLC), a suite of cognitive capabilities supporting multiple modes of learning, enabled on the mobile, and delivered through the cloud. CLC will be a both a learning and a teaching companion – an essential aid for the student as well as the teacher in the blended learning journey, one that also keeps the parents engaged in the process. An essential feature of CLC is the ability to capture fine-grained user interactions with content and device, and respond to it appropriately based on the context. These interactions will be aggregated over time and analyzed to develop rich learner models (e.g., knowledge models, learning styles) based on which actionable insights will be provided to the teacher and student in various models of learning. CLC will be designed to operate in conditions of intermittent network connectivity (common to regions like Africa and India).
As a multi-modal system, CLC can operate in school mode, remote mode, and an interactive mode, and can effortlessly switch from one mode to another based on the learning context while continuously updating knowledge, interest and interaction models as users interact with content. As it moves from one mode to the next, it elevates its functionality from being an Assistant to the teacher (school), to being a Partner to the teacher and student (remote), to being Personalized Tutor supporting adaptive learning (interactive). In each of these roles, CLC provides a suite of cognitive capabilities to assist the end user.
For example, in school mode, CLC can recommend in-classroom teaching materials for a curriculum topic, taking into account student learning activities and progress outside of class. It can infer when a student is lacking attention, and intervene as needed. It can also suggest personalized homework assignments and tutoring sessions for students. In remote mode, CLC acts as a partner bridging the inside and outside class experience, by intelligently facilitating remote discussions between the students and teachers, leveraging previously asked questions and answers in the process. It also captures contextual information associated with the remote mode which may impede the student’s learning and can advise the teacher on appropriate interventions. In interactive mode, CLC can act as a Personalized and Adaptive Tutor, intelligently selecting and sequencing tutoring concepts and learning content, adapting the experience to the student’s progression, and providing problem solving scaffolding to guide the student along the way. Across all the modes, CLC updates learner models related to knowledge, interests and learning styles, preserves learning history, and provides feedback from one mode to the others to propagate the learning context and ensure a seamless experience.
CLC will be built on top of the Learning Content Hub (LCH) and Student Information Hub (SIH) middleware components that have already been developed as part of the Watson Foundations for Education (WFE) initiative, and are being made available on Softlayer. It is thus very well-aligned with the WFE strategy of building compelling applications on a common foundational platform. In contrast to the Personalized Learning Pathways application of WFE that provides case-based intervention support to learners at-risk, CLC will focus on providing a seamless learning and teaching experience across multiple modes of learning. It will be a key exemplary application of the proposed Cognitive Learning Environment application suite of WFE. CLC will contribute to and use services from the following components of the Cognitive Research platform: Analytic and Insight Services, Knowledge and Data Services, Experience Services.