Integrated micro-learning
One of the first approaches was created for computers and it used the screensaver of the devices to detect the inactivity of the users (when switching on, when accessing a remote server, when accessing a website, etc.) and trigger training activities. A first version, defined in [24], was coined as Integrated Micro Learning (IML), where the integration of the training activities in daily life was carried out in a manner that regular activity could only be resumed after having accomplished a training micro-activity. In order to assess the level of acceptance of this system, a small study where users could refuse micro-training was carried out. However, after the pilot phase, it was concluded that 75% of the activities were accepted. Later, in 2007, another pilot study was conducted with 74 cards and 30 users. Results showed that users consumed 15 cards per day on average (in other words, 3,000 cards per year if considering that a year has a total of 200 working days), a very positive figure with regard to the potential acceptance of this training paradigm.
The evolution of this system allowed for a more elaborate deployment over Windows XP that made it possible to use more sophisticated cards such as multiple choice-only one correct or multiple choice-multiple answers. Training activities were initiated after finishing a task but before starting a new one, so the system was not exclusively focused on screensaver activation. The first pilot test to evaluate this deployment was conducted in 2010 with 62 users, who were public workers of the Austrian government. In this case, besides monitoring the records provided by the tool itself, a satisfaction survey was carried out. Results were very positive, as 91% of the users that had begun the course (82%) succeeded to complete the training, with an average consumption of 143 cards. The satisfaction survey endorsed these data since users found the experience very positive.
Micro mobile learning
Although micro-learning techniques can be used with all kinds of devices, it should be noted that portable devices (telephones, tablets, etc.) are clearly the most convenient for this new paradigm. It is easy to deduce that both paradigms go hand in hand if understanding mobile learning as “Any sort of learning that happens when the learner is not at a fixed, predetermined location, or learning that happens when the learner takes advantage of the learning opportunities offered by mobile technologies” [38]. In fact, some definitions of micro-learning explicitly mention the possibility of resorting to on-the-move learning: “Microlearning is a new research area aimed at exploring new ways of responding to the growing need for lifelong learning or learning on demand of members of our society, such as knowledge workers. It is based on the idea of developing small chunks of learning content and flexible technologies that can enable learners to access them more easily in specific moments and conditions of the day, for example during time breaks or while on the move.” [23].
Both micro-learning and mobile learning are based on reusable, autonomous, self-sufficient and linkable content. Micro-content is usually adequate for both paradigms; thus, some authors deal with micro-content design without taking into account the paradigm for which it will be employed [41]. As a consequence of these similarities between both approaches, the term Micro Mobile Learning has already been mentioned in the literature to refer to the conjunction or integration of both, being this conjunction characterized by brevity, ubiquity and interactivity [11]. Brevity, because micro-content is designed in brief blocks; ubiquity, because the training can be carried out anywhere; interactivity, because the interaction between the trainer and the student is crucial in the process.
Because of these reasons, the technological evolution led the work started in [24], described in the previous section, to a third deployment based on the use of mobile devices, co-financed by two European projects. The first pilot assessment test was carried out in 2011, with the participation of 22 public workers from different sectors of the United Arab Emirates government who were presented with a training sequence of 71 cards. In this case, users were free to use the client for mobile phones or personal computers (PC). In addition to the good monitoring results of the training, users showed their satisfaction in a follow-up survey. Nowadays, this system is distributed to different companies and administrations.
[58] addresses another aspect: the use of Micro Mobile Learning for professor training purposes. These authors identified three different models: (i) independent learning, where learners decide their own learning path (goals, how to tackle different problems, etc.); (ii) question-based learning, where trainers ask different questions to learners so that they can work towards their resolution, and (iii) collaborative learning which establishes a common aim for a group of learners who have to work together in order to achieve it. Transversely, three different interaction modes are included in each of these models: Mode 1 refers to the usage of a short message service (SMS), Mode 2 refers to an interactive consultation mode and Mode 3 refers to the use of web browsers.
A completely different alternative was proposed by [7]. In this case, interaction is not directed by the instructor and there does not exist collaborative interaction between the trainees either, regardless of the students’ needs, i.e. this approach fits better as a ubiquitous IoT (Internet of the Things). In this context, the learning process is led by the trainee’s context. These researchers, who belong to the House_n research department of the MIT (http://web.mit.edu/cron/group/house_n/), propose to place a set of sensors throughout the learner’s house to enable interaction with the learner’s mobile phone in order to trigger micro-content. When these sensors detect the smartphone and/or specific activity of the users, a micro-learning session is triggered. All kinds of sensors are employed (Fig. 4): from cards, buttons and/or RFID stickers, to movement sensors adhered to the remote control of the television. To test this system, the authors decided that the objective would be to learn new vocabulary in a foreign language (Spanish). The system was tested by a married couple for four weeks, during which time words and phrases were presented to the users with an average frequency of 57 per hour. Despite these figures, the users found the interaction interesting and effective.
The best example of success in this area is probably the one described in [11], where a micro-learning environment specifically oriented to the use of mobile devices is proposed. This environment uses the KnowledgePulseⓇ MicroLearning system (KPFootnote 1), developed by Research Studios Austria FG (RSA FG)Footnote 2, to which the authors of the publication adhere. Fig. 5 shows the commercial image of the product. This system sends small units of educational content in the form of interactive learning cards, being the content particularly suited for mobile devices such as mobile phones. Learners are asked to give a short answer, normally just true or false. Continued interaction with content generates a workflow that strengthens the acquisition of new knowledge. This learning environment is more oriented to the memorization of propositional content and uses a learning algorithm [9] [10] which is customized depending on the answers and the interaction of each student. Subsequently, the system, in its version KnowledgePulseⓇ 2.0, is also capable of supporting certain social interaction and collaboration. Thus, users can create their own cards and share them with peers.
This system is based on the Leitner system, named after the German journalist Sebastian Leitner, who devised it in the seventies. This procedure consists of the spaced repetition of short texts that need to be memorized. These texts are organized in groups or boxes. When the student acquires the new knowledge, the text changes boxes or groups, in a way that it will be presented again at a longer interval. When the knowledge has not been acquired correctly, the text goes to another box or group and is presented to the trainee again at a shorter interval.
The solution provided by KnowledgePulseⓇ MicroLearning is based on this method. The trainer has to previously divide the content into a set of learning cards, in a manner that each of them represents a different step in the memorization process. These steps will need to be sequenced according to a didactic sequence. Taking all this into account, the KP generates the repetition processes needed to guarantee the memorization of the content in the short and long terms. Besides, the KP system detects when the device (mobile phone) is not being used in order to activate a training card, the one corresponding to the user, which will depend on personal track record and on the didactic sequence of knowledge.
Composition of micro-content: an SOA approach
The composition of micro-content to elaborate or define training sequences requires a system which can provide solutions to store, to locate and to compose micro-content. The design of the solution given by [4] is based on the SOA (Service Oriented Architecture) paradigm, where communications are managed through an ESB (Enterprise Service Bus). This flexible proposal undoubtedly raises a structure in layers where the definition of enhanced metadata for the description and characterization of micro-content is particularly important. The lower layer, or Persistence layer, is intended to store the elements to be combined, that is, the micro-content (Learning Objects, LO), with the support of the metadata set that describes them. The following layer, the Semantic Layer, is made up of two broad modules that are intended to manage micro-content so that it is semantically compatible with the requirements of the Semantic Web Services (SWS). The following layer, the Semantic Web Services Layer, is intended to manage the web services which publish, discover, negotiate and combine micro-content. Finally, the authors also provide a tool to annotate the content in a simple way.
Ontological support for the description of educational content (WSMO extension) was conceived as an extension of the WSMO ontology (Web Service Modeling Ontology) [19], which was in turn extended to adapt to the specific requirements of micro-content by using, for this purpose, the LOM (Learning Object Metadata, [28]) standard. Two main difficulties were faced. First, the representation of learners to ensure that they can be characterized through their own competences and learning styles. Second, the interconnection of educational elements through a range of relations: different versions, authoring, technical requirements, etc. In fact, these relations make it possible to establish an educational sequence through the composition or sequencing of micro-content [13].
Micro-learning content in the cloud
The cloud computing paradigm fits in perfectly with the concept of micro-learning, since if the latter is characterized by the use of brief educational resources upon request, the cloud computing paradigm is also supported by the flexible use of resources (hardware, software, storage, computation, etc.) whose amount dynamically varies depending on what is needed at each particular moment. Consequently, the application of this paradigm to the provision of micro-learning environments can be considered as a natural evolution from the first platforms based on web environments, which would allow the availability of storage, backup services and computer services, elastically provided depending on the needs at each moment, and at a more affordable cost. In addition to the typical advantages of this paradigm, also applicable to most of the software and hardware solutions (cloudonomics), micro-learning systems could benefit from (i) the creation of libraries of educational micro-resources that are accessible in a simple and functional way; (ii) the use of micro e-portfolios, where the progress made by each user can be registered; (iii) the customized access to micro-resources depending on these micro e-portfolios and (iv) easier real-time interaction with other users, by using micro-blogs, micro-chats and different systems of intercommunication based on the cloud computing paradigm [35].
In [32] an architecture characterized by the following elements is proposed: (i) use of portable devices (smartphones and tablets), since they facilitate mobility, dynamism and flexibility, all of which are inherent to micro-learning; (ii) application of OCR (Optical Character Recognition) programs to process the data entries, since the use of such programs makes this task faster and simpler, thus more convenient for the learner, and (iii) application of web scraping techniques to quickly analyze the content of web pages and highlight the most relevant information. The combination of these three elements allows for a workflow where users, after accessing educational content, use software embedded in their browser (add-on) in order to add different elements of various web pages with the aid of a web scraper. Finally, this enhanced content is updated in the cloud and can be synchronized to any portable devices to access the micro-learning educational experience whenever the user deems appropriate.
Finally, cloud computing allows enhanced content to be subsequently accessed by other users, which results in a collaborative and dynamic environment that is inherent to micro-learning. With this aim, the aforementioned authors [32] show a typical micro-learning cycle (introduction, activity and conclusions) and provide a labelling system based on six categories, context, frequency, recentness, semantics, preference and feedback, which help to classify and retrieve micro-learning content.
In the same vein, the research work of [44] proposed a service-oriented system which enables the organization of a Virtual Learning Environment (VLE) for portable devices (telephones, tablets, etc.) based on collaborative learning and micro-learning. The architecture of such VLE consists of three functional modules that have been respectively implemented in order to: (i) obtain historical information about the learner (track record); (ii) capture data of the learner while interacting with the system and (iii) synchronize with the cloud. Additionally, as the most innovative contribution, the VLE incorporates two services: TaaS (Teamwork as a Service) and MLaaS (Micro-learning as a Service).
The first service, TaaS [47], deployed in Amazon EC2, combines the benefits of collaborative work with the implementation of the learning flow. The module has five web services (Survey Service, Jigsaw Service, Bulletin Service, Monitor Service and Inference Service) which should preferably be kept active and working in sequence so that all benefits can be obtained, even though they can be decoupled and adapted to the interests of the VLE. The Survey Service provides the necessary support to guarantee communications from mobile devices despite the potential instability of connections when using these devices. The Jigsaw Service makes it possible to organize discussion between learners, whereas the Bulletin Service allows assigning tasks to different work teams. As for the Monitor Service, it facilitates that all the students can monitor the work of the rest of the team members. Finally, the Inference Service is intended to assign specific tasks to each student depending on individual learning styles and preferences.
The second service (MLaaS, also implemented in Amazon EC2) is intended to provide micro-content for those users who prefer to resort to their mobile devices to take advantage of little spare moments (e.g. while commuting to and from work). The architecture shown in Fig. 6 was designed with this aim. Here, the Learner Modelling block is dedicated to characterize the learner (track record, interaction with the system), while the Real-Time Learner Data Retrieve Service is intended to obtain data on the progress and the availability of learners (in other words, when they prefer to study and the amount of time they devote to this activity). The Learning Resources Representation block is intended to manage all the possible representations of the available micro-content so that the most adequate version can be provided as needed. For this purpose, micro-content is labelled by using standard metadata for the description of micro-units [8] [52], by incorporating information such as keywords, duration, language, popularity, difficulty, etc. Finally, the Adaptive Engine uses the information that comes from these three sources or sub-modules to adapt content to each learner’s learning styles and preferences, learning context and individual needs. Therefore, this engine becomes the centre or core of the global system: it is in charge of selecting and adapting the content which is most appropriate for each micro-learning process.
This is an inclusive proposal of micro-learning in an environment based on the cloud, oriented to mobile devices and underpinned by collaborative work. The same authors continued their work by analyzing a very relevant issue when dealing with adaptive environments: cold-start problems, that is, problems to begin the adaptation when enough information to characterize the students is not available. While it is a recurrent problem in all the adaptation or recommendation environments, [45] suggested an adequate solution to the OER (Open Education Resources) that is also integrated into the global proposal as a module oriented to service and is provided in the same architecture in the cloud. The full study, with all the intermediate solutions, can be consulted in [46].
Commercial approaches
There is a wide range of options when deploying a micro-learning platform. In spite of the difficulties they must face, some proposals have proved successful in this context. The most noticeable ones, which perhaps cannot be directly associated with this paradigm, are YouTube, TED and the Kahn Academy. YouTube allows users to organize their own training when they desire to acquire some new skill or knowledge in a fast and effective way (for example, to prepare a dish with the aid of a new recipe, to meet a new origami challenge, to install a new device, etc.). Indeed, YouTube has become a source of multidisciplinary and ubiquitous knowledge, comparable with Wikipedia in terms of dissemination but used for the gradual display of content in a training context, with a very enjoyable and digestible format: video. The second of the above-mentioned proposals, TED, is based on the altruist dissemination of knowledge spread by international or local experts (in its TEDx version) through brief recorded talks that can be easily watched via its web platform. Finally, the Khan Academy is a non-profit educational organization founded by Salman Khan, who is a graduate of the MIT and the Harvard University whose main objective is to provide structured knowledge in the form of brief lessons or presentations on certain topics that the learner can freely select on the web platform. Moreover, the Khan Academy enables learners to access interactive content in order to put in practice the knowledge they have acquired. In addition to these three proposals, it is worth mentioning some other approaches:
-
The KnowledgePulseⓇ MicroLearning system http://www.videotelephony.com/knowledge-pulse-micro-learning/, which was developed by Research Studios Austria FG (RSA FGFootnote 3, as described in section 5. The system is aimed at a very specific public: trainees in corporate contexts.
-
Grovo (https://www.grovo.com/), a company founded in 2010, has offered a solution based on the SaaS (Software-as-a-Service) paradigm since 2013. This proposal allows for combining and adjusting micro-content: short videos (144 seconds), brief audio lessons and short and interactive visual elements. Trainees can combine these multi-format elements to enhance their learning experience. Since this solution is intended for corporate environments, it allows trainers and managers to use activity-monitoring tools based on modules of data analytics. Fig. 7 shows the appearance of the platform when accessed by the manager and the trainees. It has been successfully employed in large international corporations such as PepsiCo or Gap Inc.
-
The Coursmos platform (http://coursmos.com/), founded in 2013, also provides micro-content. They are normally brief lessons that take no more than three minutes, which make up over 38,600 courses that usually take no more than fifteen minutes in total. This platform was originally developed for mobile devices, especially telephones, this being the reason why it was distributed through the Apple Store and Google Play. Nevertheless, a web solution was launched in 2014. It includes a module of recommended content for learners, who are more than 1,700,000 nowadays. Additionally, Coursmos incorporates a clearly collaborative orientation allowing its users to create content. As shown in Fig. 8, there are four different Access modes, from the most basic one to those specifically intended for enterprise environments. Each of them offers a particular range of functionalities, which is reflected in their cost. Differences are mainly in terms of storage capacity, number of courses that are supported, number of events, number of authors, the possibility of customizing the domain, use of monitoring tools and applications adapted to mobile devices, etc. Nowadays, the platform is being successfully used for training purposes in more than 100 companies.
-
There are different solutions for mobile devices that propose micro-training environments, even though this term is not explicitly used. Among them, it is worth mentioning Duolingo (https://www.duolingo.com/), which is a language learning app. Students subscribe to the language they want to learn and the app gradually provides them with sequences of flashcards, in a similar way to that described in (Bruck et al., 2012), as explained in section 5 of the present document. As learners progress through the course, the flashcards that are presented to them become more and more complex. Phrases, vocabulary and grammar are transmitted through the audio and the images, without resorting to videos. Moreover, the tool has an audio recognition system to identify whether the student’s pronunciation is adequate or not. Fig. 9 shows the appearance of the application: an example of the progress made by a learner in the different headings can be seen on the left, whereas one of the tests that the student has to pass to keep making progress is shown on the right. This application incorporates gamification and socialization techniques so that the students can publish their progress in the social media and receive mentions or hallmarks, which entitle them to extra time to use the application for free.