PROCEEDINGS OF THE
14
th
ANNUAL CONFERENCE
ON WORLD WIDE WEB APPLICATIONS
7-9 November 2012Durban South Africa
Editors: A. Koch P.A. van Brakel
Publisher:
Cape Peninsula University of Technology PO Box 652 Cape Town 8000 Proceedings published at http://www.zaw3.co.za ISBN: 978-0-620-55590-6
TO WHOM IT MAY CONCERN
The full papers were refereed by a double-blind reviewing process according to South Africa’s Department of Higher Education and Training (DHET) refereeing standards. Before accepting a paper, authors were to include the corrections as stated by the peer-reviewers. Of the 72 full papers received, 64 were accepted for the Proceedings
(acceptance rate: 89%).
Papers were reviewed according to the following criteria: Relevancy of the paper to Web-based applications
Explanation of the research problem & investigative questions Quality of the literature analysis
Appropriateness of the research method(s)
Adequacy of the evidence (findings) presented in the paper Technical (e.g. language editing; reference style).
The following reviewers took part in the process of evaluating the full papers of the 14th Annual Conference on World Wide Web Applications:
Prof RA Botha
Department of Business Informatics Nelson Mandela Metropolitan University Port Elizabeth
Mr AA Buitendag
Department of Business Informatics Tshwane University of Technology Pretoria
Prof AJ Bytheway
Faculty of Informatics and Design
Cape Peninsula University of Technology Cape Town
Mr A El-Sobky Consultant
22 Sebwih El-Masry Street Nasr City, Cairo
Prof M Herselman Meraka Institute, CSIR Pretoria
Mr EL Howe
Institute of Development Management Swaziland
Dr A Koch
Department of Cooperative Education Faculty of Business
Cape Peninsula University of Technology Cape Town
Dr DI Raitt
Editor: The Electronic Library (Emerald) London
Mr PK Ramdeyal
Department of Information and Communication Technology Mangosuthu University of Technology
Durban
Prof CW Rensleigh
Department of Information and Knowledge Management University of Johannesburg Johannesburg Prof A Singh Business School University of KwaZulu-Natal Durban
Prof JS van der Walt
Department of Business Informatics Tshwane University of Technology Pretoria
Prof D van Greunen School of ICT
Nelson Mandela Metropolitan University Port Elizabeth
Further enquiries: Prof PA van Brakel
Conference Chair: Annual Conference on WWW Applications Cape Town
+27 21 469 1015 (landline) +27 82 966 0789 (mobile)
Reflections on delivering a cross-discipline, cross-cultural,
international, masters-level collaborative course using e-Learning
technologies
W.S. Leung University of Johannesburg Johannesburg South Africa wsleung@uj.ac.za D.A. Coulter University of Johannesburg Johannesburg South Africa dcoulter@uj.ac.za C.C.M. MoesTechnical University of Delft Delft
The Netherlands C.C.M.Moes@tudelft.nl
I. Horváth
Technical University of Delft Delft
The Netherlands I.Horvath@tudelft.nl
Abstract:
This paper presents a case study on the experience of delivering an Internet-based international collaborative semester course at intermediate postgraduate level and attempts to distill a model for exploring the success factors involved when presenting such courses. The pedagogic and practical implications in terms of the effectiveness of the technological and system administrative choices made for delivering the course content and facilitating team interaction is evaluated. The structure and contents of the course are considered in light of the focus on innovation within the development of novel ubiquitous products and services. Furthermore the synchronous nature of the real time collaboration tools employed are contrasted with asynchronous alternatives given that the participating countries are highly distributed according to longitude and diverse according to their native language, cultural customs and technical skill set. A survey of extant technologies employable in facilitating such courses is presented together with an analysis of the criteria necessary to facilitate a successful delivery of the course. The institutional needs of security and management of its ICS resources are placed in the context of the effective
administration of such a course given the highly heterogeneous selection of technologies employed. The contrast of the available resources of the only first world participant, functioning also as the virtual host (the Netherlands), with the remainder of the consortium is made in terms of technological and communication constraints. The problem is then modelled as a multi-agent collaborative communication problem in which participants are represented by proxy agents within a virtual environment. Multi-agency is selected as a framework for analysing the problem domain due to its strong theoretical underpinnings in computer science, economics and other fields. Each agent attempts to maximise its own utility function value based on its ability to solve sub-tasks assigned to it coupled with their contribution to the global utility for the agent’s team. Communication channels exists between agents, vary in terms of their availability over time and are not uniformly distributed between agents. Furthermore such channels may be classed as either synchronous or asynchronous. Likewise in order to model the cross-disciplinary nature of the course each task requires certain competencies to be completed effectively, these competencies are not uniformly distributed amongst the agents. To model differences in time zones each agent's cognitive capacity also varies according to a time based function and point of origin. This model is formalised using standard software engineering modelling languages and salient aspects implemented using standard agent oriented development frameworks. Although the system models a collaboration the reality of such interactions are that agents are inherently self-interested. Due to this, the model allows for differing task allocation strategies to be employed and the emergent effects of policies such as egalitarian, meritocratic and historical performance based distribution schemes to be compared. The end result of this modelling activity is to establish the most favourable mix of communication channels which would need to be provided for a given globally distributed team composition. The model may also be used to explore the most favourable policy-mix from an institution's point of view in order to facilitate such collaborations while maintaining an adequate level of security.
Keywords: Distance learning, cross-cultural e-collaboration, ubiquitous design, multi-agent modelling, security best practices
1. Introduction
A subset of distance education (Hassanzadeh et al, 2012), e-learning has been embraced by higher education as one way in which technological developments may be harnessed to meet increasing enrolment targets set by policymakers and stakeholders within the education sector (Michael, 2012). The embodiment of various information and communication technologies in which students and staff are able to transcend boundaries imposed by physical capacity, time and distance to access information and interact with one another can thus be seen as a viable supplement to the more traditional physical classrooms.
Naturally, the adoption of e-learning is not without critics raising questions over learning methods, the quality of the curricula developed, and how best to incorporate technologies
and learning techniques (Hassanzadeh et al, 2012) to provide the necessary structures for establishing successful virtual learning environments (VLEs). The offering of courses in e-learning environments require students to adopt new skill sets (Jethro et al, 2012). In essence, there is a need for a guide outlining what may be considered best practices in the offering of online classes (Michael, 2012).
Drawing on the experiences and lessons learnt from participating in a distance education programme spanning across four universities, a number of factors that played some role in influencing how the aspect of communication contributed to the degree of the programme’s success are identified. Such factors are then incorporated as considerations in our multi-agent-based modelling of cross-border e-learning environments in order to establish how best to configure the technological and learning components of a VLE to achieve maximum effectiveness.
The paper is thus structured as follows: Section 2 introduces the online course, providing an overview of its objectives in order to establish discipline, class size and year level. Details of the participants in the course are then supplied, establishing the various differences amongst the stakeholders in the latest incarnation of the programme. In Section 3, observations made during the participation of the programme are reviewed, highlighting in particular, communication-related challenges and the solutions that were employed to address such concerns. The paper concludes with a summary of the results of the multi-agent modelling and a set of identified best practices that could serve as a list of improvements to be applied in the next iteration of the programme under discussion. 1.1 Research objectives and exploratory development
The paper contributes to the growing body of experience in the design, implementation and administration of international collaborative postgraduate research in the emerging field of ubiquitous technologies. The lessons learned year serve to lay the foundation for the evolution of this course and to aid i others in the design of similar modules.
In terms of its contribution to the state of the art this work considers the following research questions.
● Is the current state of the art in online collaboration tools sufficiently advanced to support cross disciplinary, cross cultural, geographically dispersed innovative development at postgraduate level across the digital divide?
● Can the problem of coordination and collaboration across divergent skill sets serve as the basis for devising a new Collaboration Game for further exploration in the multi-agent systems research community?
In order to determine the scope of the experimental process for exploring the possibility of developing a novel variation on multi-agent collaborative games we refer to Olivier’s
discussion of prototype development in Information Technology in Figure 1. Due to the end goal not being the verification or testing of established theory the exploration branch of the workflow was selected.
Figure 1: Experimental Process Selection (Olivier, 2009)
2. Background
2.1 Philosophy and conduct of the course
Designed for widely-based, cross-cultural and multi-disciplinary international cooperation, the Ubiquitous Product and Service Innovation (UPSI) course is one answer to the challenges that have been raised by strong societal trends and demands. At its conceptualization, the current challenging mega-trends, such as globalization, well-being, and sustainability, as well as the trends of the enabling technologies, such as moving towards knowledge/innovation economies, novel computing technologies, and the evolution of the networked society, have been taken into consideration.
The main assumption of the course is that while design has been put into the position of the engine of progress and well-being at the end of the twentieth century, innovation
should be in this position at the beginning of the twenty first century. Therefore, UPSI aims at product innovation in the context of cyber-physical products and services. Towards this end, it explains the students what knowledge they need and how to generate a rich search space of feasible innovation opportunities by systematically matching combinations of technological affordances with social demands and application opportunities.
The process of innovation is presented to the students as a mechanism of putting ideas into novel useful forms and bringing them to market. The emphasis is on innovation oriented thinking and methods, rather than on application domains. In this context, innovation is being used to mean more than just ground-breaking research and technology-implied invention or development. Instead of the traditional discovery → invention→ development → product → market → profit model, UPSI introduces a new model for organizing the course contents and the scenario of the students’ exploratory, creative, and business activities. This novel process flow is shown in Figure 2.
Figure 2: Different evaluation tools
The course provides information on various existing and emerging technologies (such as sensing, mining, transmission, networking, reasoning, conversion and actuating) as well as on collaborative research and iterative innovation methodologies. The course also involves a large-scale project that synthesizes operative research, collaborative innovation, and multi-aspect prototyping. The students, product design and engineering masters, work in mixed international teams.
In order to rapidly develop the needed competencies, UPSI intends to transfer professional and methodological knowledge in various intensified forms, e.g. by invited lectures, virtual round-table sessions, and consultations with experts. In addition to enhancing the students’ attitude towards innovation, the course also increases their social sensitivity,
readiness for cooperation over geographic boundaries, ability for comprehensive abstraction and synthesis, and learning on demand.
The practical and theoretical parts of the course complement each other. The course provides information on various existing and emerging technologies (such as sensing, mining, transmission, networking, reasoning, conversion and actuating) as well as on collaborative research and iterative innovation methodologies. The course also involves a large-scale project that synthesizes operative research, collaborative innovation, and multi-aspect prototyping.
The students, product design and engineering masters, work in mixed international teams. In order to rapidly develop the needed competencies, UPSI intends to transfer professional and methodological knowledge in various intensified forms, e.g. by invited lectures, virtual round-table sessions, and consultations with experts. In addition to enhancing the students’ attitude towards innovation, the course also increases their social sensitivity, readiness for cooperation over geographic boundaries, ability for comprehensive abstraction and synthesis, and learning on demand.
The practical and theoretical parts of the course complement each other. The course decomposes to five modules, each of them pursuing different objectives and closed with a progress assessment. The module structure and the objectives of the learning and doing activities are shown in Figure 3.
Figure 3: Organizational modules of the course
In comparison with traditional design courses, elements of novelty are operative engineering/social research, business plan development, brokerage towards companies,
and concept demonstration by abstract and tangible prototyping. Furthermore, UPSI ends with a closing workshop, including a public symposium and exhibition. The students are learning not only from experts and instructors, but also from each other in socialized forms.
2.2 The second ubiquitous product and service innovation course
Presented for the second time, the UPSI course involved four universities, distributed across four continents, from Colombia, India, the Netherlands and South Africa. Four teams were formed from more than thirty Colombian, Indian and South African students, with each team being placed under a coaching team from one of the universities. The composition of each team was done with input from the coaches who best understood how to group available student disciplines in order to ensure well-rounded teams.
Running during the first half of 2012, planning for the course took place the two months prior, during which refinements were proposed and applied due to problems identified from the first UPSI programme. A schedule was set up with input from all participating universities to decide on dates as well as lecture content. Each planning session, subsequent lectures and coaches’ meetings were held by connecting to the Delft University of Technology’s video conferencing facilities, a practice that was exercised throughout the course a minimum of three times per week.
For the purpose of archiving all documentation, staff and students alike were instructed to upload all work to the Delft University of Technology’s course management Blackboard Learning System. Students were thus able to access course information and upcoming lectures in addition to submitting their drafts and final versions of project deliverables for evaluation.
Overall, the course provided us with the opportunity to observe an extensive number of lectures, presentations and meetings taking place over a variety of communication channels for a period of nearly 8 months.
We believe that the number of communications that took place over the aforementioned observation period provides us with a reasonable representation of the general challenges that participants in an e-collaborative environment can normally expect to encounter. Such communication-related challenges are presented in the following section.
3. Communication in e-collaboration
As first time participants in the programme, it soon became apparent to the South African authors that agreeing to participate in UPSI would present many wide-ranging challenges, the first and foremost of which, would be to secure the facilities necessary to meet with the
other coaches regularly via video-conferencing to discuss the programme schedule and content.
Not only was the use of potential video-conferencing facilities costly and time-restricted, but due to incompatibilities between the two systems’ hardware, it was not possible to establish a connection between South Africa and the Netherlands. Further investigations eventually led to the use of the EVO collaboration network, a web-based video conferencing system originally developed for scientific collaboration on the Large Hadron Collider project (Caltech, 2012). Figure 4 illustrates the components of the H.323 standard to provide the necessary audio-visual communication over the Internet (ITU, 2009).
Figure 4: H.323 components of a typical protocol stack (ITU, 2009)
The use of EVO proved to be greatly beneficial as it was not only free, but did not require additional special hardware to operate. This should be particularly suitable for resource-constrained institutions struggling to book video-conferencing facilities. Our Indian colleagues, for example, had to make the request for the facility well in advance. Changes to the schedule meant that they were unable to attend meetings and lectures on a number of occasions. Being developed in Java, the EVO client allowed staff members and students to connect using a variety of platforms, namely Windows, Linux and Mac OSX. EVO is of course, not without problems - during the course of the programme, the South African group experienced a number of occasions where it was not possible to use EVO to connect. At times, the fault could be attributed to the client software’s incompatibilities with the Java Runtime Environment, in particular, with the latest build of the Java Virtual Machine. More often, downtime was the result of the university’s information security policies that were in place.
In order to explain how security policies in place would impact on the use of a video-conferencing system, the communication setup of the H.323 standard needs elaborating. Primarily, the EVO network comprises two components - Koala, the client and the Panda servers to which users connect via Koala.
Figure 5: H.323 communication setup (ITU, 2009)
As seen in Figure 5, Koala clients connect indirectly with the desired video-conferencing facility. When Koala is started, it automatically attempts to connect to the first available Panda server. Due to existing security policies, Koala clients within the university were limited only to the local Panda server. On one particular occasion, the local Panda server was not functioning and queries directed to the server’s maintainers revealed that the single person in any position to rectify the problem was away for a week. Connections to EVO made outside of the university’s networks reveal that the Koala client can connect via other Panda servers, suggesting the need for additional Panda servers to be made accessible within the institution.
Such a security issue was clearly not limited only to the South African side - attempts by the South African coaches to get the other students to meet via EVO were met with zero success due to problems with the network on both the Indian and Colombian sides. Various other video and/or sound-enabled avenues were considered, ranging from Skype to Google+ Hangouts to various university-specific applications, all of which failed due either to security policies or very poor connectivity issues.
Another factor that played a role in the switch to a more text-based approach can be attributed to the inability for the students to understand each other verbally - despite repeated attempts to get students to speak more slowly to accommodate team members, students continued to struggle, prompting the need for communication to take a written form. Although the switch did improve understanding, it still resulted in the few occasions where team members were convinced that the person on the other end was shouting or angry with them. It was not until team members met in person at the final workshop that it was simply how the person expressed themselves. Interestingly, students did not appear to experience as much trouble when it came to expressing themselves through their
disciplines although this may be due to tasks being appropriately allocated to the relevant skill sets.
In the end, the inability to connect all countries for team meetings via video resulted in most teams switching to less bandwidth-intensive options such as e-mail, Google Talk, the text-based messaging feature of Skype, and Facebook, all of which were often used asynchronously.
The need for asynchronous communication was made apparent very early on in the programme with both students and coaches realising the vast time differences between the four countries - the Indian students would often struggle to stay up whereas the Colombian students would have just woken up on their side of the planet. To complicate matters further, confusion with establishing a meeting time increased when the Netherlands switched to Daylight Savings Time halfway through the course.
As the course was not the only commitment each student had, it was also soon realised that students would not always be able to attend their meetings with their team leaders or attend all scheduled lectures. In particular, one team was unable to reach an agreed upon time in which all team members could meet, resulting in one or more team member being absent at each meeting.
As can be surmised, several communications-related factors have been noted to influence a team’s ability to collaborate effectively in a virtual environment:
● Connectivity - the ability to exchange ideas and work with fellow collaborators can vary due widely due to the availability of numerous synchronous and asynchronous options. Connectivity can also be impeded by existing information security policies and unstable software or infrastructure.
● Language & Culture - a lack of understanding of how other cultures and customs work can lead to incorrectly perceived hostility, leading to aggravated members that do not wish to work with one another.
● Time zones - the geographically dispersed nature of the team members meant the possibility of key figures missing during decision-making.
● Discipline / Skill set - although impacting on e-collaborative efforts to a lesser degree, incorrect work allocation had the potential to delay submission or decrease the quality of project deliverables.
4. Multi-agent analysis
The UPSI collaboration problem, which was informally discussed in previous sections is dominated by the interactions of independent entities acting to realise both their own individual goals and in order to achieve collective success for their team as a whole. Systems such as these are dominated by the emergent effects of the interactions between these independent entities. Highly decoupled, highly concurrent systems with such characteristics are amenable to modelling as multi-agent systems.
From a computer science or artificial intelligence point of view an agent is a programming abstraction which evolved from traditional object oriented programming. Two of the key principles in object oriented programming are those of information Encapsulation in which a set of related state values for an entity are unified within a single data structure together with the operations which are permissible on. While this work will focus on the concept of agents as they pertain to modern computer system design the idea itself has existed in the literature of numerous disciplines from long before the era of modern computing. Initially agents and agency served primarily as an abstraction tool which has been employed in the analysis of the emergent properties of complex systems. Two such classical disciplines are Economics and Law both of which have made use of subtly different realisations of the concept of agency that capture two distinct, salient aspects of the concept which strongly relate to the UPSI collaboration problem.
Modern economic theory makes use of agents as typically self interested decision makers and economic actors within a model. These agents may play the roles of various real world entities such as buyers or sellers in a highly granular or small scale market simulation or of entire countries or industries in highly rarefied or abstracted macroeconomic models. While these agents initially only served as conceptual building blocks in the overall model construction they were adopted by Game Theoreticians as computational units (game theory describing the decision making process as it does in purely algorithmic terms) (Magliocca et al, 2011). Modern computational economic models often create explicit programmatic implementations of these agents blurring the distinction between the theoretical economic construct and modern programming paradigm.
The legal and philosophical concept of an agent occurs as part of the agent-principal problem. In this case an agent is an entity that acts on behalf of another. In law the distinction in terms of responsibility between the principal (the action originator) and the agent (the action executor) is minimized and is understood in terms of the Latin maxim “The one acts through another, acts themselves” (qui facit per alium, facit per se). This presents a problem in terms of determining ultimate liability in the domain of artificially intelligent agents acting on behalf of a human principle. As Figure 6 indicates the key problem is introduced due the asymmetrical distribution of information and opportunities for action.
For the purposes of this paper the authors would like to draw a parallel between the self interested agents depicted in the classical agent-principal problem and participants in the UPSI collaboration problem. In both cases each participating agent is individually self interested and self motivated yet all are collectively judged by an external party on the collective quality of their efforts. The collaboration problem is in fact a many-to-many extension of the agent-principal problem.
4.1 Agent-wise decomposition of the collaboration communication problem
In order to explore the emergent properties of the UPSI collaboration problem and establish it as a novel variation on existing communication thought problems the following implementation was created using the Java Agent Development Framework (Bellifemine et al, 2007). Agents in JADE are created via inheritance and composition. All agents must extend the base jade Agent class and then fill their own behaviour pool container with classes created by extending the base behaviour class. All agents initially run a short one shot behaviour instance which serves the role of bootstrapping initialisation and service discovery activities within the agent to allow agents to discover each other and be discovered using the directory facilitation service within the JADE platform.
The agent system developed is heterogeneous in the sense that there are two main species of agent. CourseManager agents contain a single instance of a cyclic behaviour child class which continuously responds to messages sent to it from the Collaborator agents. The CourseManager agent essentially fills the role of the project mentor by informing each Collaborator agent of the composition of its team as well assigning work to the group as a whole.
Inter-agent communication is achieved via the agent communication language message performative types together with serialized java byte code objects embedded within the message. This allows for the message typing of speech act theory together with the flexibility of using high level Java constructs the types of which would otherwise be lost due to the type erasure inherent in Java’s generics implementation
Assignments are represented as a vector of task objects. Each task is suited to completion by an agent possessing an appropriate Skill which are instances drawn from a Java enumeration.
Collaboration proceeds as follows:
● Agents are initialized to have a collection of random links to other agents in the system. Participation in a given task is limited to those agents which are connected according to the graph topology of the communication channels. All agents are linked to the course management system.
● Agents once instantiated make contact with the CourseManager (i.e. log on to the course management system) .
● Agents are informed of the composition of their groups and assigned tasks.
● In order to complete a task the agent polls those agents to which it is connected via the direct edges in the communication topology graph.
● Those agents each contribute their skill towards completing the assigned task. For the sake of simplicity it was assumed that each task was equally difficult. If no agents in the local subgraph contain the skill required in order to complete the task the agents inform the CourseManager of this and the task is returned to the pool.
● The CourseManager tracks the success rate of agents together with the types of communication channels linking it to to its neighbours.
The simulation software was run on a variety of problem sizes and team compositions. While it would be folly to attempt to draw overly general conclusions from a very specific abstraction of a real world problem it is worth noting that the presence of asynchronous communication channels (particularly with the CMS which were always present) played a large role in allowing tasks to be completed. While it is unlikely that such software would be used in planning the delivery of such a course the mindfulness which results in considering the scope of the problem cannot be overstated. The UPSI communication problem was established as an interesting problem and is worthy of further study in agent oriented system development.
5. Conclusions and future work
As established from the generated results of our agent modelling, communication is generally best maximised through asynchronous means. This is in line with the best practices identified from our observations made during the course. As distance learning is aimed at reducing constraints, it makes sense that an approach which avoids confining and restricting the working hours of both student and staff would be more beneficial to the overall productivity levels of the teams.
Lectures, for example, could be recorded and made available to students for access at a later stage. Such a facility will enable students who either missed the lecture due to some unforeseen circumstances or struggle with English to go over the lecture again in their own time.
It should be noted that the paper does not suggest the complete abolition of synchronous communication as regular meetings amongst team members and coaches play a vital role in establishing relationships that are necessarily for students to work well together at the final workshop. Given the short time to assemble and test the prototype, students must already know how to work with one another, cognisant of each other’s strengths and weaknesses prior to the start of the workshop.
From an information security perspective, there is one conclusion that can be made to maintain a relatively reliable and cost-effective video-conferencing facility via the EVO network - as discussed in Section 3, the current nature of Koala does not permit users to manually specify a Panda server at startup. Institutions that permit only a single Panda server are thus courting potential disaster should that particular Panda server fail. Our recommendation is therefore to suggest that information security departments add an additional number of Panda servers distributed worldwide to the institution’s whitelist so that Koala clients are able to connect to the EVO network even in the event of the default Panda server being temporarily inoperational.
Typical of a novel e-learning course, the UPSI programme is one that is constantly being revised and improved in order to engage students and inspire them to develop truly innovative prototypes. With a third run of the UPSI programme being tentatively scheduled for September this year, we would like to test our results by applying the suggested favourable communication channels mix as generated from our agent modelling. The planned inclusion of additional universities in the collaborative programme will enable us to evaluate our results on a larger scale.
6. References
Fabio Bellifemine, Giovanni Caire, Dominic Greenwood. 2007. Developing Multi-Agent Systems with JADE. ISBN: 978-0470057476 John Wiley & Sons Ltd.
Caltech. 2012. EVO FAQ. Available: http://evo-wiki.cern.ch/twiki/bin/view/FAQ/WebHome (accessed 29 July 2012)
Hassanzadeh, A., Kanaani , F. & Elahi, S. 2012. A model for measuring e-learning systems success in universities. Expert Systems with Applications, 39(12): 10959-10966. International Telecommunications Union. 2009. Packet-based multimedia communications systems. ITU-T / Publications / Recommendations / H Series / H.323. Available http://www.itu.int/rec/T-REC-H.323/en/ (accessed 29 July 2012)
Jethro, O, Grace, A, & Thomas, A 2012, E-Learning and Its Effects on Teaching and Learning in a Global Age, International Journal Of Academic Research In Business & Social Sciences, 2(1):203-210.
Magliocca, N., Safirova, E., McConnell, V. & Walls, M. 2011. An economic agent-based model of coupled housing and land markets (CHALMS). Computers, Environment and Urban Systems, 35(3):183-191.
Michael, K. 2012. Virtual classroom: reflections of online learning. Campus-Wide Information Systems, 29(3):156 - 165.
Olivier M, Information Technology Research - A Practical Guide for Computer Science and Informatics, Third edition, Van Schaik, 2009, ISBN 9780627027581 (184 pp)
Acknowledgements
The South African authors of this paper would like to thank Professor Elize Ehlers for giving them the opportunity to take part in the UPSI programme and to Dr Ian Ellefsen for lending his technical knowledge in the setup of the necessary facilities required for the running of the video-conference lectures.
