How Micro Learning will Transform Learners Engagement

Learners must decide their learning agenda by choosing the courses they should take to continue to enhance their knowledge. However, asking learners to decide their learning agenda places an undue burden on the learner. After all, learners don’t know what they don’t know! It takes experience to set goals, to monitor and reflect on the learning progress, and to keep oneself continuously motivated. In addition to being forced to choose courses without knowing the content, learners can only customize their learning agenda at the level of the course instead of being able to match learning content to their specific learning needs.

While self-paced online learning comes with the advantage of time and location flexibility, the downside is the requirement of self-regulated learning skills. Once the learning has been deconstructed into modular LOs  prescribed by a set of attributes, they can be matched with the needs and constraints of learners in an emergent fashion, without the need for learners to proactively choose learning content. This automated approach to recommendations is widely used in social networks, entertainment, and online marketplaces. More than 80 percent of TV shows people watch on Netflix are discovered through the platform’s recommendation system. Netflix uses machine learning and algorithms to find shows that they might not have proactively chosen. The system consists of three elements: Netflix subscribers who consume TV shows and rate their enjoyment of TV shows; Netflix “taggers” who label each TV show on a set of attributes; and machine learning algorithms that match content with subscribers. We can adapt this recommendation engine approach to the lifelong learning context by developing a “Netflix for Learning.” The automated learning recommendation system would consist of four elements: The user-defined in terms of her needs and constraints, the recommendation engine that captures the user behavior and uses this knowledge to retrieve relevant LOs from the repository, and the LO aggregator engine, a set of AI-based natural language processing and generation and visual cognition algorithms which automatically stack the LOs to high-level mini-modules structures to create an ontology for the learning topic. However, just stacking these objects together is not enough.

Du et al. (2008) emphasize that the usage of LO as a teaching vehicle is only meaningful if it is embedded in an appropriate context. Building a storyline has a crucial impact on knowledge absorption. Automating the process of creating a customized learning journey requires a semantic understanding of the LOs and the needs of the learner. On both sides, data is available in structured and unstructured formats, which requires AI-based technologies to process them. Again, the innovations are already present. Recently the Goethe-University in Frankfurt, Germany, developed an AI-based approach to scan 52,000 scientific documents of a specific topic and to automatically create a readable literature review, which was published under the name of the algorithm: Beta Writer. The process consisted of several steps of data retrieval, data preprocessing, Data structuring, and content aggregation as visualized in Figure 3.

‍

‍

Drawing inspiration from the problem of using machine learning based technologies to create an automatic literature review, we propose the following steps for creating a customized learning agenda from individual learning objects:

Data Retrieval

The recommendation engine matches the learning objects to the learner’s needs. Learning objects are tagged and categorized based on hard measures (content, applicability for specific job roles, industries, costs, ratings, or seniority levels) as well as soft measures (teaching style, voice style.

Structuring

The characteristics of a LO (metadata and cognitive characteristics) are used to create a topic map. A topic map describes the knowledge of a specific domain and links it to existing information resources. Topic modeling algorithms help to extract semantic information from structured and unstructured data such as documents, images, or videos. The topic map is the basis of the learning storyline. It is used to create a meaningful flow of learning objects. As a result, the LOs can be hierarchically and sequentially structured into mini-modules (Figure 4.)

Generation

The mini-modules are wrapped into baskets of introductory content, main content, summarizing content building a storyline or learning plan. Such a design, using a beginning, middle, and end, provides the necessary context to foster the knowledge absorption. A set of learning plans will be compiled by ranking the different mini-modules within each basket of the story. LOs might come in different shapes, such as images, text or video, and different length using various examples and from different institutes. A page rank algorithm can be used to pick the mini-modules with the LOs that best fit the learner’s need for each basket of the storyline.

Excerpt from Prof Mohanbir Sawhney Future of Learning : https://www.mohanbirsawhney.com/blog

‍