Open corpus architecture for personalised ubiquitous e-learning

0 downloads 0 Views 308KB Size Report
Sep 26, 2007 - people have access to Internet-enabled devices such as desktop PCs ... the access network that enables a fast, reliable and per- ... and Internet delivered courses are playing a more central ... factor that determines the value of the e-learning for an ... Model (SCORM) and Extensible Markup Language.
Pers Ubiquit Comput (2009) 13:197–205 DOI 10.1007/s00779-007-0189-5

ORIGINAL ARTICLE

Open corpus architecture for personalised ubiquitous e-learning Cristina Hava Muntean Æ Gabriel-Miro Muntean

Received: 4 December 2006 / Accepted: 8 May 2007 / Published online: 26 September 2007  Springer-Verlag London Limited 2007

Abstract As the e-learning area matures, there are a growing number of e-learning content providers that produce and distribute material that covers a large range of topics, differs in quality and is represented in various formats. Lately, different devices and various network technologies allow extensive user access to educational content almost anywhere, anytime and from any device. Ubiquitous e-learning has the potential to provide continuous and context-based, educational material to human learners anytime, anywhere and on any device. Since each person has different expectations related to the content, the performance of the delivery and display of that content, it is desirable for an ubiquitous e-learning environment to provide user-oriented personalisation of e-learning material. However very often there are multiple sources of e-learning material at various web locations (open corpus resources) that cover the same topic, but differ in terms of quality, formatting and even cost. It is very difficult for learners to select the content that best suits their interests and goals, characteristics of the device used and delivery network as well as their cost budget. This paper proposes an innovative ubiquitous e-learning environment called Performance-based E-learning Adaptive Cost-efficient Open Corpus frameworK (PEACOCK) that provides support for the selection and distribution of personalised

C. H. Muntean (&) School of Informatics, National College of Ireland, Mayor Street, Dublin, Ireland e-mail: [email protected] G.-M. Muntean (&) Performance Engineering Laboratory, School of Electronic Engineering, Dublin City University, Dublin, Ireland e-mail: [email protected]

e-learning rich media content (e.g. multimedia, pictures, graphics and text) to e-learners such as it will best suit users’ interests and goals, meet their formatting preferences and cost constraints, while considering the limitations introduced by the end-user devices and the delivery networks to the user. PEACOCK’s main goal is to maximise the users’ e-learning experience and increase their learning satisfaction and learning outcome. Keywords Adaptive e-learning systems  User-oriented personalisation  Cost-efficiency  Ubiquitous e-learning environment

1 Introduction Educational content is delivered via many electronic media including network-oriented (e.g. web), broadcast (e.g. digital and interactive TV) and package-based media (e.g. CDROM, DVD, etc.). However, the network-oriented approach—also known as web-based e-learning or simply e-learning in this paper—has attracted the most interest and lately has gone through a period of rapid growth. This exponential growth is due to the high demand for educational content mainly by academic and industrial users, but also by learners from other categories that span from rapid information seekers to life-long distance learners. Currently most interest in web-based e-learning is from university and corporate training users, but as the newest technology become more affordable and widespread, an increase in the percentage of the other types of users is expected [1]. As the e-learning area matures and a large number of people have access to Internet-enabled devices such as desktop PCs, laptops, PDAs, e-learning service providers

123

198

are becoming more interested in the delivery of media-rich content over a variety of networks, as attractive as possible and at a low cost. Lately, the variety of wireless technologies such as WiFi, WiMax and 3G mobile networks launched on the market are making e-learning more attractive to the users by offering access to content anytime and anywhere. In this context the learning environment does not need to be designed for a particular device or for a certain type of network connectivity. The environment constantly follows the learner regardless of the location the learning takes place. These latest technological advancements are pushing the e-learning field towards a new stage that provides support for the development of a ubiquitous e-learning environment. The ultimate goal of ubiquitous computing is to provide seamless and unobstructive computer-based services to users regardless of their device. In this context the latest devices such as i-mate Jasjar Pocket PC for example enable users to have access to multiple networks at the same time from a single device (Fig. 1). These networks may differ in characteristics such as bandwidth, level of congestion, mobility support and cost of transmission. In these new ubiquitous e-learning environments it is equally important to provide learning material that best suits user learning goal and display device and to choose the access network that enables a fast, reliable and performant transfer of the e-learning content at a low cost that matches user budget constraints. This overall cost includes both the price paid for the selected educational material and the delivery cost. Adaptive e-learning systems seek to make the e-learning content more attractive by tailoring it to individual users goals and interests. However, all these benefits are lost when the user’s operational environment, the network or device through which they access the online content, cannot support the delivery of personalised e-learning

Fig. 1 User choice between multiple access networks

123

Pers Ubiquit Comput (2009) 13:197–205

material. Moreover if the e-learner cannot or does not want to pay for the selected material the efforts of the personalisation process are not appreciated. In consequence there is a need for a system that would bring the user-oriented tailoring of e-learning material to a higher level of personalisation: apart from focusing on learner’s interest and goals, it should also look at the performance of delivery and display as well as at the cost of accessing the learning material. This paper proposes a novel innovative ubiquitous elearning solution denoted Performance-based E-learning Adaptive Cost-efficient Open Corpus frameworK (PEACOCK) that provides support for the best content selection and remote delivery of personalised distributed media content to e-learners. PEACOCK’s goal is to optimise users: – – – – – – –

learning goals and interests, favourite media content formatting, costs for content purchase and its delivery, while taking into account: users’ personal characteristics (knowledge, goal, interest, etc.), device display and processing power limitations, delivery network constraints and users’ budgets.

PEACOCK’s architecture, described generically in Fig. 2, provides the meeting place where e-learning content providers and e-learners can exchange high quality services. PEACOCK supports high level of personalisation in its selection of e-learning material based on learners’

Fig. 2 Performance-based e-learning adaptive cost-efficient open corpus framework

Pers Ubiquit Comput (2009) 13:197–205

interests and goals, their favourite media formatting, access network and device capabilities while being also aware of existing users’ budget limitations. The paper is structured as follows. The next section presents major trends in e-learning. They include customisation and personalisation of content, rich media formatting, delivery over heterogeneous wired and wireless networks, quest for interoperability and need for performance and cost-based adaptation. Section 3 summarises existing major adaptive e-learning environments and highlights both their characteristics and achievements. Section 4 describes in details the proposed PEACOCK and presents its potential benefits. Section 5 draws conclusions.

2 E-learning trends The web-based e-learning market is divided into two main sectors: formal education and corporate training. The rate of adoption of online learning in the education sector is growing, therefore the quest to create courses and lessons that can be reused and combined in modules is an attractive one. It is estimated for example that distance education will account for 50% of all post-secondary learning by 2010 and Internet delivered courses are playing a more central role in distance education or in supporting conventional delivery methods [2]. It is also estimated that 20% or more of corporate training is now being conducted online [1]. One reason for this spurt in growth has been e-learning ability to link to other enterprise systems. Another driver of demand has been a shift towards greater use of ‘‘self-service’’ systems. Apart of the high demand for e-learning systems, there is a growing interest in the customisation and personalisation of the content delivery. IDC has indicated that the level of customisation of content is the most important factor that determines the value of the e-learning for an organisation and there is a strong trend towards customised content. A wider access to broadband, WiFi and 3G mobile networks is making e-learning even more attractive for both users and companies. Most e-learning is now accessed over the Internet, but companies such as InteractiveServices [3] and Ossidian [4] have started to deliver e-learning content over mobile networks. Also multiple formats (e.g. text, image, audio, video) for encoding e-learning content and methods for delivering it (e.g. file transfer, streaming) are currently used in both education and corporate training sectors. Therefore it is expected that video-based and character-based e-learning content production and delivery over heterogeneous networks will experience a significant growth in the following years [1].

199

E-learning systems interoperability and content reusability are also important factors that drive the e-learning market. As organisations make significant investments in digital learning content, they seek greater assurances of portability and reusability. High support for standards has further expanded the market for e-learning products by offering the opportunity to integrate content and applications from different vendors. So far, the most effective standards have been Shareable Content Object Reference Model (SCORM) and Extensible Markup Language (XML). Nowadays more and more content vendors claim some level of compliance or conformance to the SCORM standard to support online learning and it seems that this is slowly getting to a winning position. In this context it can be concluded that the main factors driving the e-learning field towards ubiquitous learning environments are the following: – – – – –

Growth in use of e-learning in both the education and corporate training sector. Growth in interest in personalisation of content delivery. Wider access to broadband, WiFi and 3G mobile networks. Use of multiple formats including multimedia for delivering e-learning. Greater support for standards that ensure interoperability and reusability.

As e-learning is a growing area more and more companies and educational institutes provide online e-learning solutions. In this context it is likely that an oversupply of information will occur. As it is expected, that it will differ in terms of formatting, size, quality, etc. Wider choice for material selection gives high opportunities for everyone to make use of most personally relevant content. Unfortunately the benefits of open corpus resources come with its price: information overload. It will be very difficult for e-learners to select the best content from multiple e-learning systems that matches their interests and goals as well as their network connection and device (display size, processing power, battery life). If on top of this cost aspects are added to each potential piece of e-learning information, an intelligent solution is needed in order to optimise the selection process and PEACOCK—proposed in this paper—responds to these needs.

3 Adaptive e-learning environments Researchers from both industry and academia were interested in finding solutions to deliver personalised educational content tailored to individuals or groups based on user characteristics such as skills, goals, knowledge and

123

200

preferences in order to improve the e-learners’ overall learning outcome and their performance. The adaptive elearning approach involves gathering some initial information about the user, monitoring user interactions with the system, building a user profile and adapting the delivered content to this profile. Among research-proposed adaptive e-learning systems are AHA! [5, 6], QuizPACK [7], ELMART II [8], JointZone [9], APeLS [10]. These systems build a model of the goals, knowledge and preferences of each individual person and use this model throughout the interaction with the user in order to propose content and link adaptations, which would best suit e-learners. Lately, researchers started to integrate learning styles in the design of an AHS along with the classic learner’s features. Several systems providing adaptation to users’ learning styles have been created such as INSPIRE [11] and AES-CS [12]. Tracking the user behaviour in real-time in order to retrieve an appropriate and fine-grained user profile, as well as to provide personalised learning content, represents a challenging task for the adaptive e-learning area. Apart of navigational behaviour, and page scrolling, real-time eyetracking and content-tracking techniques have been recently introduced and applied within the AdeLE [13] project and they can help to identify areas of understanding difficulty and to provide selective additional information or explanation. The main goal is to observe users’ learning activities in real-time by monitoring a number of behavioural aspects and personal traits such as objects and areas of focus, time spent on objects, the sequence in which learning content is processed, momentary states (e.g. tiredness [14]). Lately, the new wireless technologies (e.g. WiFi, 3G) have started to be used in supporting learning communities. These technologies support access to educational content anywhere and anytime. Various works in the literature have proposed solutions to support e-learning through wireless technology (ubiquitous e-learning). MUSEX [15] supports children learning and interacting in a museum employing PDAs. Lancaster GUIDE system [16] distributes information about monuments or ongoing events. The information is selected by the system according to a user’s interests and location and is displayed on wireless-enabled devices. Virtual University [17] is a learning environment at the University of Hagen, that supports distance education on mobile devices. The system distributes didactic resources and services that can reasonably be used under the given environment conditions of the learner. In parallel with the academic research that led to an important number of adaptive e-learning systems, many companies have started to produce and comercialise similar systems. IBM has launched the Workplace Collaborative Learning v2.5 [18] that recommends specific training for a student based on profiles, skills and competencies.

123

Pers Ubiquit Comput (2009) 13:197–205

Companies such as SkillSoft [19], Ossidian [4], PulseLearning [20] and InteractiveServices [3] are already developing e-learning systems that deliver content tailored to either user interests or devices over wired and wireless networks. The large majority of web-based adaptive e-learning systems are stand-alone systems dealing with limited localised resources. They are so-called closed corpus systems and use proprietary data repositories. These systems provide no support to share their e-learning resources with other systems and neither to access information stored in other locations. Nowadays web plays a key role in information access and dissemination and very important are issues such as peer review, validity and quality of information. These issues are addressed to a limited extent by digital educational repositories (DER) that allow for safe storage, delivery, reuse and sharing of information. This information is represented in form of learning objects (LOs), where a LO is a self-contained reusable digital resource tagged with metadata and represents a small unit of learning material that supports a more complex learning activity [21]. Open corpus systems rely on a number of distributed DERs and on the reuse of existing LOs. They make economic sense, improve efficiency, allow for sharing of expertise and provide up-to-date and accurate e-learning resources. There is significant worldwide activity in both the public and private sectors in the research and development of repositories of LOs that are searchable, interoperable and accessible. The rationale behind DERs is to reduce the significant cost of developing and customising educational material. MERLOT [22] is a large public and free LO repository co-operative. Some private firms that develop LO repositories and tools to use them include SmartForce [23] and SkillSoft [19]. In addition to LO repositories there are many learning resource gateways which offer both free and non-free educational material. System interoperability and content reusability issues have been also addressed with a number of communication protocols (e.g. Open Hypermedia Protocol), guidelines and standards for the representation of resources (e.g. SCORM, AICC, IMS). Since many producers of e-learning material can easily add LOs to DERs, oversupply of information may occur. Therefore, personalised support for e-learners becomes important, when e-learning takes place in open and dynamic learning and information environment. Open corpus adaptive educational hypermedia systems (OAEHS) are adaptive e-learning systems that implement advanced teaching strategies and content adaptation techniques, and enable integration of existing learning material distributed over the web. These systems use DERs and deliver personalised content according to user needs regardless of the

Pers Ubiquit Comput (2009) 13:197–205

location of the content. For example, the delivered material can be part of an university tutorial or a company LO. Several OAEHS (e.g. ELENA [24]) were proposed in the academic community. However, the advantages brought by these systems are lost when the users’ operational environment, the network or device through which they access the selected content, cannot support the delivery of personalised e-learning material. Moreover if the e-learner does not want to pay for a whole course or a large material, the efforts of the personalisation process are not appreciated. In consequence, the selection of distributed LOs should be based not only on users’ characteristics but also on the connectivity and cost properties in order to allow for a cost-efficient fast transfer from the source to user terminal.

4 Performance-based e-learning adaptive cost-efficient open corpus framework The PEACOCK provides support for offering a complex set of e-learning-related services to both learners and content providers that exchange e-learning material. PEACOCK’s main purpose is to allow for the selection, personalisation and distribution of e-learning rich media content (e.g. multimedia, pictures, graphics and text) to elearners such as they will best suit users’ interests and goals, meet their formatting preferences and cost constraints, while considering the limitations introduced by the end-user devices and the delivery networks to the user. Its main goal is to maximise the users’ e-learning experience and increase their learning satisfaction by selecting and providing rich media e-learning content that best matches users’ expectations given existing cost, device and network constraints. Figure 3 presents PEACOCK block-level architecture. It consists of two main components:

201





PEACOCK e-learning management system (PELMS), localised centralised and deployed on a server widely accessible by e-learners, represents the brains of the system and PEACOCK DER, distributed on the network at different locations, often third-party owned, provide e-learning material to be selected and distributed to users.

Another actor in PEACOCK’s architecture is the User that avails from e-learning services offered by PEACOCK by accessing PELMS from its regular client browser. Following that the user receives information tailored to their characteristics, cost preferences, access network capabilities and device properties.

4.1 PEACOCK’s e-learning management system PEACOCK’s PELMS is a complex three-tier server-side system that allows for interactions with both users and DERs and provides PEACOCK’s adaptation and personalisation mechanisms. The outermost tier allows users to access PELMS services through its e-learning interface (eLI) that provides presentation-related adaptability. This ensures that the e-learning content is best displayed on each device type. eLI also protects PELMS from potential harmful contacts. The middle-tier includes the main modules of PELMS that support its functionality. They are e-learning adaptation module (EAM), DER management module, performance monitor, billing module and statistics module. E-Learning Adaptation Module is one of the most important modules as it is in charge with the whole process of selection of personalised content that best suits the learner’s characteristics, learning flow, access network, device properties and cost limitations. This adaptation is

Fig. 3 Performance-based elearning adaptive cost-efficient open corpus framework’s blocklevel architecture

123

202

Pers Ubiquit Comput (2009) 13:197–205

based on building of an user profile that represents these learner-related properties and on the functionality of an Adaptation Engine that collects the best content that matches certain user profile in current learning context. –







The DER management module manages DERs distributed across the network, supporting services such as new DER registration with PELMS, concept definition and navigation between concepts based on their relationships, retrieval of distributed LOs, etc. Performance monitor monitors different performance metrics such as end-to-end response time, throughput, QoS metrics, etc. in real-time during user interaction with the system. Performance monitoring is required in order to provide feedback to the EAM in order to improve its adaptation process by building a more accurate user profile. Billing module is in charge with computing and recording billing information in relation with the cost of the content accessed by the learners. The billing will provide support for both e-learners, as they will pay for the learning material they used and the content providers, as they will be paid for the LOs they own that were accessed by users. Statistics module aims at providing information following the functionality of the PELMS that would help both the content providers to improve the quality and the popularity of the LOs or concepts that they offer and the learners to be informed. The information statistics module provides is in terms of mostly accessed LO format from various devices, mostly used end-user devices, average length of an e-learning session, average user population characteristics that have accessed certain content, etc.



4.2 PEACOCK’s digital educational repositories The DERs are very important components of PEACOCK as they store the LOs, concept hierarchy and include a representation of the relationship between concepts and LOs. In general they belong to certain content providers that manage them completely. These content providers can also use their LOs independent from the PEACOCK (for example as part of their own learning management system), attracting other revenues if they want to. Figure 3 presents the main components of a DER at block-level. They are: educational material database, concept model and concept authoring tool. –



The innermost-tier includes a set of databases that support information storage for the PELMS modules. They are concept database, DER info database, user profile database and billing database. –





Concept database stores in centralised manner a brief description of each concept. In order to ensure the information consistency across all DERs, a concept will be used by a DER only if it is registered with PELMS. Registration of new concepts is permitted only if there is not an already registered concept that best matches the description provided. DER info database stores information about each DER registered with PELMS via the DER management module. This information includes ownership and location (such as IP address, port number). User profile database is the repository where the AM saves the user profiles. It is consulted at every new access to provide the latest information about learner-

123

related characteristics and therefore to allow AM to best tailor the content. It is automatically updated after each access based on the information gathered during the delivery of learning material. Billing database stores both learner and content provider billing-related information such as accounts, addresses, bills, billing items, discounts, etc.



Educational material database stores DER’s LOs. PEACOCK’s requirement is that these LOs are SCORM compliant in order to allow for their reuse and management. LOs could represent multimedia, audio, graphics, images and/or text and may have various formats displayable and playable by widely available client browsers and their plug-ins. Concept model stores the concept hierarchy that represents the relationship between various concepts and associates concepts and LOs. Figure 3 presents how abstract concepts are hierarchically related at the level of each DER and how leave-concepts have associated concrete LOs. It also indicates how the same concept (e.g. A) could have different relationships with other concepts and could be associated with different LOs in various DERs. Those LOs allow for alternative selection by PELMS based on specific learner characteristics. Concept authoring tool is the application that allows the content provider to access the PELMS in order to register its DER and DER located LOs with the PEACOCK system. This tool will also allow for the registration of new concepts with PELMS and usage of already registered ones, the creation of the concept hierarchy as presented in Fig. 4 and the association between concepts and SCORM-compatible LOs. Each LO will have associated a SCORM-like tag that will describe their topic (for content-related selection), ownership (for copyright), cost (for billing), format and size (for displaying on certain devices and

Pers Ubiquit Comput (2009) 13:197–205

203

Fig. 4 Concept hierarchy and relationship with LO collections at various DERs

transmission in some network conditions). Apart from the original registration, each DER is required to inform PELMS if any changes of the LOs or in the concept hierarchy have occurred. PEACOCK assumes that the LOs are already created. QoE-LAOS [25, 26]—an extension of the classic LAOS authoring model [27] is used for the specification of the concepts’ attributes, links that may exist between concepts, as well as relationships between concepts and LO. A set of algebraic operators divided in four categories: constructors (e.g. create, edit), destructors (e.g. delete), visualisation (e.g. list, view, check) and compositors (e.g. repeat) was defined in order to create and manipulate the concepts or links. QoE-LAOS also associates metadata that address delivery and display-related performance issues to each concept that has a physical representation in the form of a text, image, multimedia clip or any combination of them. Metadata is an abstract representation of the most significant features (e.g. size, resolution, average bitrate, frame rate, etc.) that characterise these physical instances of the concepts and affect in any way their delivery or display performance. More details on description, formalisation and exemplification of the QoE-LAOS can be found in [26].

4.3 PEACOCK’s e-learning adaptation module One of the most important components of PEACOCK’s PELMS is the EAM that is in charge with the cost-efficient selection and distribution of e-learning content. When provided with a user e-learning goal, EAM’s main purpose is to process existing information stored in PELMS’s databases, to make content selection from the LOs available in distributed DERs, to suggest certain access network and to deliver the selected content to the user. The EAM performs content selection based on the user goal and on a user profile that is built for each user and is updated at every user-system iteration. The user profile, apart from user characteristics and interests,

includes information about budget, available access networks, their price-plan and their current traffic status. As maintaining an accurate user profile is very important for the accuracy of the e-learning adaptation, building the user profile involves two phases. At registration PELMS gathers information about user device, access networks and their status. The system detects some of these features, whereas the user explicitly indicates some other characteristics that cannot be automatically discovered. Then PELMS monitors continually the process of content delivery to the user and, if changes occur in some of the network, device or user-related characteristics, it informs the EAM which updates the user profile. User profiles for all the PEACOCK users are stored in a user model. The EAM is defined as a collection of condition action (CA) rules applied/assessed at every user access and of event condition action (ECA) rules that are triggered by events. ECA events indicate changes in either device properties or network-related performance characteristics and can happen anytime during web session, including during the transmission of a multimedia stream. CA rules follow an IF-THEN format while ECA rules the WHENIF-THEN format. These adaptation rules are used to determine which information (concepts) will be presented to the user or what modifications of one or more features of the content to be delivered to user should be performed in order to suit user device characteristics and/or network properties. For this, the rules make use of the information on user profile (user model), on concepts’ features (concept model), and on device and network conditions (performance model). These rules have associated actions to be taken that may involve insertion/removal of LOs, web link adaptations such as link sorting or removal, bitrate or frame rate modifications, picture resolution adjustment, etc. Next examples of CA and ECA rules are presented. IF (device resolution = 320 · 240) THEN {picture resolution = 320 · 240} WHEN (loss rate [ 2) IF (bandwidth \ 1) THEN {bitrate, 0.384; framerate = 8}

123

204

More details on adaptation rules formalisation and various features of the content, device and networks that can be combined in the adaptation rules can be found in [25, 26]. Authoring in relation to the EAM aims to formally describe the adaptation rules according to either an ad hoc scheme or to a pre-defined adaptation algorithm. Regardless of the approach chosen, it is very difficult to involve automation in this process as it is highly dependent on what information the authors choose to consider on feature sets (e.g. bitrate, resolution) that describe a particular content (e.g. multimedia or image) and which are their optimal values. It also depends on factors (e.g. loss, bandwidth) that have an impact on the performance of content delivery and/ or display. Currently the authors must construct (or select from various existing solutions) a suitable adaptation algorithm and translate this algorithm to the format required by EAM’s adaptation rules. However, potential performance or QoE-aware adaptation algorithms do exist, such as one that adjusts static web content and was proposed in [28] and the Quality Oriented Adaptation Scheme (QOAS) that adjusts multimedia content and was described in [29]. It is also important to note that there is a tension between adaptation for user interests (which might indicate the inclusion of content such as high resolution video) and adaptation based on network connectivity and device (which might indicate that the content can only be delivered at a lower resolution or cannot be delivered). Currently, the author has to be aware of such potential conflicts and to resolve them. This task is made more difficult as the e-learning delivery engine may choose to apply the adaptation rules in different orders (or even iteratively) and may even choose to deliver different content.

5 Conclusions This paper proposes and presents PEACOCK, an innovative ubiquitous performance-aware e-learning environment that provides a very efficient meeting place between content providers and e-learners. The former provide e-learning material and have an extra source of revenue, whereas the latter access content. PEACOCK enables intelligent selection and remote delivery of personalised media content to e-learners such as to optimise their learning goals and interests, favourite media content formatting and delivery costs. This is achieved while considering existing device display and processing power limitations, delivery network constraints and user budget. The ultimate goal of PEACOCK is to maximise the users’ e-learning experience and increase their learning satisfaction by selecting and providing rich media

123

Pers Ubiquit Comput (2009) 13:197–205

e-learning content that best matches users’ expectations given existing cost, device and network constraints. The solution proposed by PEACOCK is simple and effective. As it relies on third-party owned and managed repositories of e-learning material, it can easily build an exponentially growing library of content that will attract e-learners and thus further e-learning content providers, etc. The choice of content based on user budget will also make it attractive to various e-learner types. On top of this the adaptation based on network and device capabilities brings the e-learning service provision to a higher level of personalisation to users’ characteristics, closer to e-learners’ expectations.

References 1. Bersin J (2004) E-learning research year in review and predictions for 2005. In: What works in e-learning newsletter. Bersin & Associates, Oakland 2. Stephen Abram (2003) A primer on e-learning...The framework, the market, the players, vol 12, issue 2. KMWorld Magazine, San Jose 3. Interactive Services, http://www.interactiveservices.com/ 4. Ossidian, http://www.ossidian.com/ 5. De Bra P, Calvi L (1998) AHA! An open adaptive hypermedia architecture. New Rev Hypermedia Multimedia J 4:115–139 6. De Bra P (2002) Adaptive educational hypermedia on the web. Commun ACM J 45(5):60–61 7. Brusilovsky P, Sosnovsky S, Shcherbinina O (2005) User modeling in a distributed e-learning architecture. In: Ardissono L, Brna P, Mitrovic A (eds) Lecture notes in artificial intelligence 3538. Springer, Berlin, pp 387–391 8. Weber G, Brusilovsky P (2001) ELM-ART: an adaptive versatile system for web based instruction. Artif Intell Educ J, Special Issue on Adaptive and Intelligent Web-based Educational Systems 12(4):351–384 9. Ng M, Hall W, Maier P, Armstrong R (2002) The application and evaluation of adaptive hypermedia techniques in web-based medical education. Assoc Learn Technol J 10(3):19–40 10. Conlan O, Wade V, Bruen C, Gargan M (2002) Multi-model, metadata driven approach to adaptive hypermedia services for personalized e-learning. In: Proceedings of second international conference on adaptive hypermedia and adaptive web-based systems (AH 2002), Malaga, Spain, pp100–111 11. Papanikolaou KA, Grigoriadou M, Kornilakis H, Magoulas GD (2003) Personalizing the interaction in a web-based educational hypermedia system: the case of INSPIRE. User Model User Adapt Interact J 13(3):213–267 12. Triantafillou E, Pomportsis A, Georgiadou E (2002) AESCS: adaptive educational system base on cognitive styles. In: Proceedings of the 2nd international conference on adaptive hypermedia and adaptive web based systems (AH’2002), workshop on adaptive systems for web-based education, Malaga, Spain, pp10–20 13. Gu¨tl C, Pivec M, Trummer C, Garcı´a-Barrios VM, Mo¨dritscher F, Pripfl J, Umgeher M (2005) AdeLE (Adaptive e-Learning with Eye-Tracking): theoretical background, system architecture and application scenarios. EURODL II, September, http://www. eurodl.org/materials/contrib/2005/Christian_Gutl.htm 14. Pivec M, Pripfl J, Trummer C (2005) Adaptable e-learning by means of real-time eye-tracking. In: Kommers P, Richards G (eds) Proceedings of world conference on educational

Pers Ubiquit Comput (2009) 13:197–205

15.

16. 17.

18. 19. 20. 21. 22. 23. 24.

multimedia, hypermedia and telecommunications, Montreal, Canada, pp 4037–4041 Yatani K, Sugimoto M, Kusunoki F (2004) Musex: a system for supporting children’s collaborative learning in a museum with PDAs. In: Proceedings of the wireless and mobile technologies in education, 2nd IEEE Intl. Workshop on Wireless and Mobile Technologies in Education (WMTE’04), pp109–113 Lancaster University, the guide project: overview http://www. guide.lancs.ac.uk/overview.html Becking D, Betermieux S, Bomsdorf B, Feldmann B, Heuel E, Langer P, Schlageter G (2005) How to match mobile learning resources with learners current needs: the didactic profiling, proceedings of HCI 2005, Las Vegas IBM Workplace Collaborative Learning v2.5, http://www.lotus. com/lotus/offering3.nsf SkillSoft, http://www.skillsoft.com PulseLearning, http://www.pulselearning.com IEEE 1484.12.1-2002 (2002) Draft Standard for Learning Object Metadata, IEEE Learning Technology Standards Committee MERLOT—Multimedia Educational Resource for Online Learning and Teaching, http://www.merlot.org SmartForce, http://www.smartforce.com Dolog P, Nejdl W (2003) Personalisation in ELENA: how to cope with personalisation in distributed e-learning networks. SINN03 Conference, Oldenburg, Germany, www.elena-project.org

205 25. Muntean CH, Muntean G-M, McManis J, Cristea AI (2006) Authoring model for quality of experience-aware adaptive hypermedia systems. In: Proceedings of the 4th international conference on adaptive hypermedia and adaptive web-based systems (AH2006), A3EH: the 4th international workshop on authoring of adaptive and adaptable educational hypermedia, Dublin, Ireland, pp 325–334 26. Muntean CH, Muntean G-M, McManis J, Cristea AI (2007) Quality of experience—LAOS: create once, use many, use anywhere. Int J Learn Technol, Special issue on Authoring Adaptive and Adaptable Hypermedia, 2007 27. Cristea AI, Mooij A (2003) LAOS: layered WWW AHS authoring model and their corresponding algebraic operators. In: Proceedings of the W3C Int. World Wide Web Conference (WWW’03), Alternate Track on Education, Budapest, Hungary 28. Muntean CH, McManis J (2006) Fine grained content-based adaptation mechanism for providing high end-user quality of experience with adaptive hypermedia systems. In: Proceedings of the W3C Int. World Wide Web Conference, Edinburgh, UK, pp 53–62 29. Muntean G-M (2006) Efficient delivery of multimedia streams over broadband networks using QOAS. IEEE Trans Broadcast J 52(2):230–235

123